The Development Assistance Committee (DAC) of the OECD is the major provider of information on funding agency aid flows. It is, however, dependent on the quality of data provided by individual agencies. The DAC holds two sources of information on the sectoral breakdown of funding agency programmes. The first, the Creditor Reporting System (CRS) is based on a comprehensive list of sector and subsector codes. In the case of education, the latest version of the CRS codes includes four general codes (not specific to level of education), three covering basic education , two for secondary education and two for post-secondary education. One limitation of the most recent structure of codes is that those who may be interested in assessing what level of support is provided to non-formal (NFE), as opposed to formal education, are frustrated by the fact that formal and NFE are 'bundled' together under each of the three basic education categories. A further difficulty facing agencies is that projects and programmes are constrained to a single sector code, which means that where an intervention spans several sub-sectors this sub-sector information is lost. Otherwise, the CRS system provides a reasonable basis for assessing the nature of funding agency support to basic education - in theory. Unfortunately, in practice there are significant omissions in the data provided by individual DAC agencies through the Creditor Reporting System.

The amounts recorded in the CRS data are substantially lower than the other main source of DAC data on basic education (see below), mainly due to the fact that several major agencies do not apply the CRS codes to the technical co-operation parts of their programmes. This results in very sizeable underreporting of basic education, which limits its usefulness if one is attempting to prepare an accurate and comprehensive assessment of basic education spend. Furthermore, for many agencies, a very large share of the aid to education is reported under the first general category of CRS education codes. This includes support to education policy and management and teacher training, for instance, much of it may actually have been destined to assist basic or primary education. It is not however, captured on the basic education totals. Related to this is the issue of support to sector wide approaches (SWAps) or sector investment programmes. The current guidelines for CRS reporting state that agencies should record commitments to such sector reform programmes under the education - level unspecified categories, such as education policy and administrative management. This is due to the difficulty of attributing what share of sector wide approaches can legitimately be deemed to be supporting basic education. However, many current so-called sector wide programmes are in fact sub-sector programmes in support of primary education. This also contributes to the under representation of basic education support under the CRS system. It is for these reasons, and in particular the incomplete coverage of CRS data, that CRS data have not been used at all in this analysis.

The second source of DAC data on basic education is provided in the DAC's Aid by Major Purposes table, henceforth referred to as the DAC Sector Table. This shows education and basic education aid flows to be far higher than that indicated by the CRS data. However, there are unfortunately major limitations to this source also. Firstly, there are some gaps in the data for basic education, though this has improved in the last couple of years. Secondly, and even more seriously, only commitments data are provided. In other words, there is no way of telling from the DAC Sector Table how much aid to basic education has actually been disbursed in a given year by a said agency. Commitments data often tend to vary sharply from year to year, partly as new political decisions are made. This can result in a significant increase in one year, followed by a steep decline in the following year, partly because the commitment recorded may in fact refer to aid that is likely to be spent over several years. Some agencies, such as DFID, supply data to the DAC Sector Table based on what is termed a 'coefficient'. This represents an educated guess as to what share of the agencies education aid is likely to have been spent on basic education. However, it may very well be inaccurate, and may not be based on a careful analysis of an agency's actual portfolio of interventions. In addition, the DAC Sector Table distinguishes between education and basic education only, shedding no light on the allocation to the many sub-sectors that make up each.

In this Section we use the original data submitted by agencies for use in the DAC Sector Table, which is referred to simply as the DAC data. These differ in certain years for certain agencies from the percentages shown in the published DAC Sector Tables. However, they provide the totals directly supplied by agencies to the DAC in US $ for education and basic education and they may be more accurate than the percentages of total aid indicated in the DAC Sector Table. They nonetheless suffer from exactly the same limitations as those highlighted above for the DAC Sector Table data.

It is a matter of concern that the standard, internationally available data on basic education are inadequate to the task of providing an accurate and comprehensive picture of aid commitments and actual disbursements to basic education, despite the significant political priority accorded to basic education at Jomtien.

It was for these reasons that the study team were asked to attempt to collect new primary data direct from the funding agencies, which sought to break down both commitments and disbursements by education sub-sectors. The process of requesting agencies to supply data and following up with them individually has been time-consuming. It is somewhat disappointing, therefore, that many significant gaps remain in the ODI survey data (see Table 5.2). Several of the agencies explained the difficulties of giving a complete account of their country's aid to education in the developing world (see Box 5.1).

Box 5.1

Definitional Issues such as where to put early childhood education (ASDB), capacity building in non-education sectors (NZODA), changes in codes during the decade (Netherlands), basket finding (DANIDA) and because the definition of basic education is left to partner governments (DFID).

Agency Capacity because allocation and expenditure data is not tracked that way (EU, NORAD), projects cover more than one sector (DFID) and to disentangle training provision takes resources and time (NZODA) especially towards end of budget year (USAID)

It is believed that where agencies have provided data directly to ODI this is likely to be the most accurate available. Significant weight is given to this data, therefore, in this Section. However, as indicated above, there are some major gaps in this data, with several large bilateral and multilateral agencies having been unable to provide data to complete the ODI survey. For this reason, the discussion below covers both the ODI survey data and the DAC Sector Table information (especially as the two sets of figures can be very different if they originate from different sources). To avoid confusion the latter figure has been presented separately in appendix 3. This is an unsatisfactory situation, which highlights the considerable difficulties agencies are facing in presenting a comprehensive and accurate account of their commitments and expenditures on basic education. We would argue that where data is available from both sources, that the ODI survey data should be preferred. The DAC Sector Table, however, can be usefully used where the ODI survey data is incomplete.

Data were requested on both commitments and disbursements broken down by sub-sector and by geographical region from 1991 to 1999. Several major agencies (see footnote below), were unable to supply any comprehensive data on basic education commitments or disbursements. Others provided either one or the other. Only a few were able to indicate which countries or regions benefited and these have been considered in Chapter 6. The data provided were converted to US dollars to allow for comparability across agencies. Trends over time for groups of agencies were calculated by aggregating volumes across agencies before calculating percentages (thus taking account of the relative size of the different agencies), rather than averaging the percentage shifts across the agencies (which would mean that each agency counted equally).

Despite the limitations of commitments data described above, this is the main focus of this Section for two reasons. Firstly, the majority of the data provided by agencies for the ODI survey were commitments rather than disbursements. Secondly, in order to compare ODI survey data with those reported by the DAC there is no option but to use commitments. When considering the trend of support to basic education over time revealed by this data, the distortion due to the intrinsic 'spikyness' or unevenness of commitments data must be kept in mind. Thus for instance, total ODA commitments for USA stood at $19.7 billion in 1990, falling to about $11bn. in 1991 and remaining around the $8.5bn. mark until 1994. This huge and one-off boost to US (and therefore total DAC) commitment levels in 1990 will inevitably affect the trend of basic education commitments expressed as a percentage of total commitments. US total ODA disbursements, on the other hand, remained broadly constant, but these figures cannot be used since neither the DAC data nor sufficient ODI survey data exist for disbursements. More fundamentally, while commitment levels are the best and swiftest indicators of changes of political priority (i.e. the weight given in funding terms to particular sectors), they say little directly about the levels of support actually received by partner countries. We have included a small sub-section on disbursements.

The figures generated by our survey for Denmark, Finland, Germany, and New Zealand correspond roughly to those received directly from DAC. The survey data show a consistently higher percentage of total aid in support of education for Canada, and lower levels for Belgium (and the figures for Ireland and for UK from the two sources hardly overlap). Note also that the DAC data supplied directly to ODI differ from those given by Bennell and Furlong (1998), who observed a decrease from 10.2% in 1989/90 to 8.5% in 1991-92 with an increase only back to the original level of 10.1% in 1993/4.

These aggregate figures, of course, mask large variations among individual bilateral agencies. Neither the survey data nor the DAC data, however show a consistent trend for the decade as a whole. According to the DAC data, although the percentage of ODA was higher at the end of the decade than at the beginning for Australia, Denmark, Finland, Germany, Luxembourg and Norway, only Luxembourg consistently (except for a small dip in 1996) increased the percentage of its ODA allocated to the education sector (for the years available), whilst there were declines for Belgium, France, Japan, Netherlands, Spain, Switzerland and the UK.

The overall share among the multilaterals for which we have data, and especially with respect to the World Bank, has increased (see tables 5.4 A and B). We received information from the Asian Development Bank, the Caribbean Development Bank, the Inter-American Bank, UNDP, UNESCO, UNICEF and the World Bank, but the data that we have received from the Caribbean Development Bank and UNDP was not sufficiently detailed to include in the tables. Further, as only the Inter-American Development Bank and the World Bank provided figures for total lending, we are only able to calculate shares of all lending for those two agencies.

However, for this limited sample, we have calculated shares of commitments to education in all ODA and basic education as a share of education aid and of all ODA, based both on all loans (Table 5.4A) and only on 'soft' loans (Table 5.4B). It can be seen that, as a percentage of all ODA, the World Bank has increased the share allocated to education from 7.2% to 10.9% over the decade, although there were dips in both 1992 and 1997, and that the Inter-American Development Bank has increased the share from 0.4% to 2.9%, although there were much higher percentages shares in 1994 and 1997. All the multilateral agencies who sent data have assigned a substantial fraction of their commitments to education to the basic education sub-sector.

One might reasonably argue that, from the point of view of children in the recipient country, what matters is the total volume of educational aid that they receive rather than necessarily the proportion of the agency's overall aid budget. Moreover, given the fungibility of aid between sectors, as well as within them (e.g. within education), the total volume of aid going to education is likely to be important, as well as the proportion of aid within the education total which is allocated to basic education.

According to our survey, the absolute value of commitments to the education sector has grown from about $400m in 1990 to nearly $1,400m in 1994 and back again to $900m in 1998 (all in current prices: the 1999 figure is based on so few agencies that it should be ignored). However, it must be noted that several major agencies are excluded from this total. The total value of disbursements, reported by five of the agencies has been highly erratic with $600m in 1991, but never more than $300m in any subsequent year.

The absolute value of commitments to the education sector reported via DAC has dropped slightly from $5,664m in 1990 to $4,946m in 1991, and stayed at about that level until 1995 when there was another increase to $6,037m, since when commitments have dropped back to $4,793m in 1997 (current prices). The recent decline mirrors the steady fall in ODA commitments overall (as we have seen that the percentage of ODA to education has continued to rise according to both sources).

Box 5.3 shows a clear upward trend in bilateral aid to education over the decade, based on ODI survey data (using whichever of commitment or disbursement data is available). Although the DAC data in appendix 3 suggest a more mixed picture, for those agencies where there is information from both sources, we believe that the survey is more reliable.

From the multilateral agencies World Bank lending for education increased from $1,487m in 1990 (all loans) to $2,252m in 1991 but has fallen back to $1,706m in 1996. In 1990, the absolute amount of Bank lending for education was slightly over 20% of all bilateral funding and by 1994 this had doubled to 39%. It is noticeable, however, that IDA loans have hardly increased at all, remaining around $800 million; although it was also true that the relative proportion of IDA lending to education to grants from the bilateral agencies doubled from 8% in 1988-89 (comparing $329m with $4,300m) to 16% in 1994 (comparing $800m with $5,000m).

During 1990-96, total lending for education among the three big Regional Banks averaged $622 million per annum approximately one third the annual level of the World Bank. Overall, using whichever of commitment or disbursement data is available according to the survey data, multilateral commitments stood at $1,000 in 1990, rising to nearly $2,000m in 1994 (a peak year), but have since fallen back to $1,300m in 1998.

It has actually proved quite difficult to extract data on resource commitments to basic education (or even to primary education) despite the prominence given to Jomtien and its re-iteration by the agencies in new or renewed policy commitments. The EFA Forum Secretariat, as part of the EFA Mid-Decade Review, noted that "only a few agency countries have declined to join the international consensus in support of basic education" but were unable to collect any data - except from Germany - to prove the point. Nonetheless, as will be seen below, there is evidence that an important and positive shift, from the point of view of Jomtien, has indeed taken place for many agencies.

Based on DAC published data, Bennell and Furlong (1998) showed substantial increases in the proportions allocated to basic (although often from a very low base) for Australia, Canada, France, Germany, Japan, Netherlands and Switzerland. However, they also showed that the response during the first half of the 1990s was relatively slow. Whilst some agencies (notably Germany, Netherlands, United Kingdom) undertook formal comprehensive reviews of their education aid programmes soon after Jomtien (see BMZ, 1992, The Netherlands (DGIS) 1994, ODA 1993), others such as Australia, Canada, France, Japan and Norway did not complete their reviews until much later. Bennell and Furlong suggested that this might mean that there would be substantial improvement in the second half of the 1990s; but also hedged their bets by suggesting that there were a number of other factors which came into play.

Based on our survey data (see Table 5.5), the proportion of aid to education which is going to basic education has been graphed for each country and aggregated to all bilaterals and all multilaterals (see Fig 5.3). For all bilaterals together, commitments increased substantially as a proportion of all commitments to education to about 43% in 1996, falling for 1997 and 1998 and reviving again in 1999 to nearly 30% (although the last percentage figure is based on only a very small number of agencies). Based on DAC data, the proportion of all education aid committed to basic education by bilateral agencies has risen steadily and dramatically from around 2% in 1993 to 14% in 1997 (Table 5.5A). Similar tables showing the aid to basic education as a percentage of all commitments are shown in Table 5.6 (ODI Survey) and in the annex for the DAC data (Table 5.6A)

For the five agencies providing disbursement data, they increased very rapidly to peak in 1995 at over 50% with a substantial drop in 1996 and rise again to over 40%. For all multilaterals together, based on our survey, the proportion of basic education lending in all education lending has been high throughout the 1990s (at least according to their definitions) at between 75% and 100%, but disbursements have remained between 30% and 50%.

Based on data from DAC publications, Bennell and Furlong (1998:55) say that, "at or around the time of Jomtien, only four bilateral agencies (Denmark, Norway, Sweden and the United States) committed and/or disbursed more than 25% of their education sector activities on basic education activities" . However, based on their replies to the King and Carr-Hill questionnaire in 1991, this was also true for Germany and the Netherlands.

Whilst there are variations between the different agencies, there has been more of a consistent pattern over the decade than was the case for the share to the whole education sector. According to the ODI survey data, four agencies have generally increased the share of commitments to basic education as a proportion of total education commitments (Denmark, Germany, New Zealand and UK), three show stability, and one a decline (see Box 5.4). According to the DAC data (see Box 5.4A in appendix 3), there are clear rises in the proportion of education commitments which go to basic education for Australia, Belgium, Denmark, New Zealand, UNESCO Extra Budgetary and the World Bank; and clear rises in the proportion of disbursements for education which go to basic education for the Netherlands, Norway, and UNESCO Extra Budgetary. Note that there are differences in interpretations for the two sources; notably for Denmark, Netherlands (partly because of differences between commitments and disbursements, partly because of different time-spans, and partly of course because the data are different).

The following types of trends in bilateral agency support since 1990 can be distinguished:

Significant increases in the commitment to basic education with at least a 20% increase in funding share to basic education. This group includes 5 countries (possibly several more but we do not have any data for the beginning of the decade).

Moderate or minor increases either from a very low base or to a small fraction (less than 0.5%). This group included 4 countries.

Declines in the share of basic/primary education. For three of four agencies that, prior to 1990 had the best records in support of basic education, recorded declines in the share of basic education

According to our survey data, the absolute amounts committed to basic education at the beginning of the decade were very small according to either source. For the group of agencies which have provided data, this rose to about $400 m in 1995 but dropped back to under $300m in 1998; and according to the DAC data there was a similar pattern rising to nearly $650m in 1995. In contrast, there does appear to have been a steady rise in disbursements from almost zero in 1990 to about $170m in 1998. Among the multilaterals, commitments have increased from $500m to a peak of nearly $2,000m in 1994 and between $1,200m and $1,900m in succeeding years.

ODI survey data (summarised in Box 5.5 below) show that the absolute value of agency commitments (or disbursements where available) to basic education has increased for the great majority of agencies (nine agencies). It declined for none, and remained more or less constant for a further three.

"The survey suggested that many agencies found difficulty in reporting their contribution to basic education. In some cases…this was because agencies collected data by different geographical and income categories, but for the main 'Jomtien components' of basic education (early childhood education, primary schooling, adult literacy etc.), it was plain that many agencies simply could not provide these breakdowns."

This was the conclusion of King and Carr-Hill (1991:15) after a similar exercise over eight years ago: it still rings true today. They went on to say:

"It was equally plain that many agencies were determined to rectify the situation", citing the responses of Switzerland and USA. Unfortunately, this does not seem to have happened: it has to be seen as astonishing that the situation has not improved in the interim. There seems little point in having a target, if one is unable to accurately reflect progress towards (or away) from it.

On the basis of commitments data, the amounts allocated to the education sector have fallen in line with the commitments overall (but see the caveat above about this trend). On the other hand, one positive element stands out vis-a-vis Jomtien: the proportion of all commitments to basic education has risen as a proportion of commitments to the whole education sector; so that the allocations to basic education as a proportion of all bilateral commitments have also risen; and the volumes of aid to the basic education sector has been stable or risen slightly over the decade.

This positive conclusion needs to be tempered with a recognition that whilst many agencies have increased the proportion of their aid to basic education in line with their policy statements, in some/many cases, this was from a very low base (less than 0.5% of their whole aid budget); and in some cases, it is still less than 0.5% at the end of the decade.

There is a major problem in reporting sector breakdowns of disbursements in general and of reporting the amounts of aid allocated/committed to basic education in particular. The situation has not improved over the decade. Moreover difficulties related to the current reporting systems are likely to be exacerbated as more agencies move to provide budgetary support to the education sector under sector approaches (see Section 10), as allocations to the basic education sub-sector will not be specified.

This raises the possibility that it is not only impossible, but also not necessarily relevant, to ensure full intra-sectoral accountability for education, since the potential strength of such budgetary sectoral support lies in part in the fact that all sub-sectors are covered. A general move by the agency community to sector wide approaches could change the basis on which reporting takes place. However, a standardised approach by agencies to these issues would still be of great value.