Tracking aid flows for development assistance for health

Abstract

Background and Objectives

The global architecture for providing development assistance for health (DAH) has become increasing complex in the last decade, with many new funding agencies entering the health sector. This study presents a detailed picture of European Union (EU) and EU member state originating DAH between 2006 and 2009; with a specific focus on assessing the extent of complementarity of development assistance sourced from the EU.

Design

We use a combination of internal EU reporting systems, OECD-DAC creditor reporting system data and other data sources to estimate DAH flows. Our method uses a line by line project assessment in order to identify and categorise DAH flows.

Results and conclusions

Our findings show a complex picture of DAH flows – from source, to channel of assistance, to channel of implementation – that is hard to track at the global level, and rarely comprehensively and regularly tracked at the country level. While the majority of EU DAH is focused on low and lower middle income countries there also remains much disparity between countries; and further analysis is required to better understand whether these imbalances are fair and efficient; or result in overlap. We also recommend investment in quality control of DAH tracking internally within donor agencies, and investment in the development of country based systems in order to enable countries and development partners better harmonise DAH flows.

The global architecture for providing development assistance for health (DAH) has become increasing complex in the last decade, with many new funding agencies entering the health sector (1, 2). Although, in part, these new agencies have been established to streamline funding, the changing architecture of DAH has raised concerns that DAH is becoming increasingly fragmented (3–5), potentially creating duplication and making the process of efficient resource allocation at the country level challenging (6, 7). These concerns need to be seen in the context of the global agreements that govern aid such as the Paris Declaration, the Accra Agenda for Action and the Busan Partnership (8). These agreements lay the foundations for increased co-ordination in DAH and founded a number of core principles, including that donors work to streamline their efforts in-country (harmonisation) and for donors and recipients alike to be jointly responsible for achieving these goals (mutual accountability) (9).

As a supranational entity, the European Union (EU), plays a key part in this architecture complementing the development efforts of EU Member States – not merely as a 28th donor, but as a coordinator, convenor and policy maker. In an effort to improve the harmonisation of aid, in 2007 the EU produced an EU Code of Conduct on Complementarity and Division of Labour in Development Policy, which provides a framework for the implementation of development assistance between the EU and EU member states, including the pursuit of more effective collaboration within the multilateral (ML) aid system.

The mapping of DAH is a necessary pre-requisite for complementary and effective programming of development assistance, and harmonization both at the global and national level (10–12). It provides the basic information necessary for EU donors managing development assistance to plan their financing in a way that maximises the joint effectiveness of development assistance; and allows them to identify and reduce the duplication of effort. It can highlight where many there may be a surfeit of donors operating, and areas that may be being ignored. It can be used globally to help EU donors decide which countries and sectors to operate in, and nationally to help planners identify resource gaps. It also provides an essential element of the accountability required by constituent EU populations.

However, despite this clear need, until recently there was a dearth of comprehensive data tracking of development assistance to the health beyond the formalised reporting on expenditures by the Organisation of Economic Cooperation and Development – Development Assistance Committee (OECD-DAC). In recent years, new notable efforts have emerged: primarily the work conducted by the Institute for Health Metrics and Evaluation (IHME) and Action for Global Health, supported by the Bill and Melinda Gates Foundation (1).

This paper aims to complement and extend these efforts by presenting a detailed picture of EU and EU member state originating DAH between 2006 and 2009; with a specific focus on assessing the extent of complementarity of development assistance sourced from the EU. We also aim to supplement and inform the methods used by others tracking aid more globally, by tracking expenditures from EU registered NGOs; using additional methods to capture multi-sector expenditures; and finally through conducting four country based studies to help identify models for future country focused tracking systems.

Methods

The measurement of DAH is complex, and methods are continually evolving (13–17). We therefore present here a detailed description of our methods in order to support those conducting such work in the future, and to allow for comparison with other studies.

Data sources

The data for this study was obtained from a variety of different sources. We used internal financial reporting systems to estimate the amount of aid sourced and channelled by the EU. We were provided with a raw dataset of all EU projects. In addition we had direct access to internal project management records, financial systems and documents: the CRIS system – Common External Relations Information System of the European Commission. For non- EU expenditures we used primarily used the OECD-DAC creditor reporting system (OECD-CRS). There have been concerns about the comprehensiveness of reporting by the various aid agencies reporting to OECD-CRS. In order to check the completeness of the OECD-CRS system, we compared the total amounts reported by the DAC annual aggregates database to the amounts in the OECD-CRS (coverage ratio). This comparison (and some further checks with bilateral sources) revealed that bilateral reporting was sufficiently comprehensive during the period 2006–9.

For MLs, our initial review of the OECD dataset revealed some gaps – and therefore in some cases we used other sources of development assistance reporting. For the health sector we accessed the dataset on the subject produced by the IHME (1). We included IHME data for those MLs who did not report any data to OECD-CRS (Asian Development Bank, Pan-American Health Organisation). IHME also provides estimates of aid from US based NGOs and these were also included in our dataset. For non-OECD-DAC members we sourced our data primarily from the AidData database.1 This database tracks expenditure by scanning project tenders. It does not capture all non-OECD aid flows, most notably development assistance from China. While we identified some data sources on aid from China (e.g. the China statistical yearbook), these could not be validated and were therefore excluded from our estimates.

We also included some data on European NGO expenditure. We were not able to find one comprehensive source for European development NGOs, so we therefore limited our work in this area to an ‘NGO scoping exercise’. We drew up a list of 30 NGOs in each sector across the EU Member States, consulting/contacting NGO funding bodies and a variety of development professionals from several member states. We then searched NGO websites and official publications to identify any health and education disbursements. A key challenge in reporting NGOs development assistance is double counting – so where possible we excluded any financing reported by the channels of assistance in the OECD-CRS dataset.

We were also not able to fully include collective EU research expenditure that aims primarily at promoting the economic development and welfare of developing countries – although we did include that provided by the European Commission Directorate General for Research and Innovation. The IHME dataset provides data on several US-based research funding agencies, such as the Bill and Melinda Gates Foundation – and this was included in our estimates.

Identifying DAH

The task of identifying the proportion of aid that goes to the health sector is complex. Most of the above data sources use OECD-DAC purpose codes (18) to classify development assistance into different sectors. These codes identify the ‘specific area of the recipient's economic or social structure the transfer is intended to foster’. This approach means that development assistance is not allocated to sectors on the basis of the types of goods, services or outputs produced by a project, nor does it refer to the character of development that a project may influence, but instead reflects the institutional structure of the recipient's country health sector. Beyond this, the OECD-DAC guidelines suggest that, when there is a choice between codes, the code should be as specific as possible. For example, the OECD-DAC guidelines suggest that a project such as the ‘construction of a tuberculosis clinic’ should be coded under ‘infectious disease control’ rather than ‘basic health service infrastructure’. The type of disease is therefore deemed more important than aspects such as construction or training, for instance. Finally, each project record includes a project title and a descriptive text, which provide additional details on the specific focus of project expenditure, which helps when classifying expenditure. Moreover, previous attempts to identify health and education sector development assistance have shown that there are high numbers of straightforward coding errors in the OECD-CRS database (16).

Taking into account the above issues, and our available time and resources, we took the following approach to clean and categorise the data from each data source to arrive at our final estimates for total DAH. The European Commission's CRIS database contains OECD-DAC purpose codes for all its projects. All projects with a health sector DAC code were selected (projects that focused on censuses were removed). We then checked each project to ensure that the title was relevant to the health sector and, if there was doubt, was checked on the CRIS database using available project documents. We then carried out a keyword search on all projects that did not have a health code. The search was conducted in four languages: English, French, Portuguese and Spanish. Lastly, we searched available project documents on the CRIS database and included them in the dataset if they were clearly health projects, using a number of decision rules (available from authors on request). While it was possible to search and review the European Commission's CRIS database and AidData datasets line by line, the OECD-CRS dataset is much larger and this approach was not feasible. We therefore first created an extract of all projects with a health or education purpose code, using a DAC code filter, and then a key word search. A summary of the filtered codes and keywords is available from the authors on request.

Finally, where projects bridged sectors we classified them into two types: 1) multi-purpose – where all funds are used simultaneously to achieve multiple objectives (e.g. training for journalists in HIV awareness). In this case we allocated the total funds to health; and, 2) multi-sector – where a project contains funds for several distinct activities in different sectors. Here we reviewed the CRIS system to identify the proportion of expenditure by sector, and only included the relevant proportion in health and education totals. For the OECD we divided the total project amount by the number of sectors referred to in the project title or description.

Identifying DAH sourced in the EU

Development assistance can either be reported by the source or channel of assistance. Typical sources of development assistance include the national treasuries of EU Member States or NGOs. The EU is also considered to be a source of financing – one with its own resources and budgetary authority. Funds from these sources are then allocated to different development agencies which are referred to as channels of assistance. These agencies fund projects and programmes in low- and middle-income countries. These include a range of bilateral government agencies, development funds, ML agencies and NGOs.

The OECD-CRS database and most other data sources report development assistance by channel of assistance for each development agency. To estimate DAH sourced in the EU some extra steps therefore need to be taken. For bilateral development assistance identifying the financing source is straightforward because they are fully financed by their respective governments (It should be noted that in this regard that by its sui generis nature the EU is to be considered a bilateral agency2). However, for ML agencies the picture is more complex. The financing of multi-lateral agencies by EU member states can be divided into two types: core and non-core funding. Non-core financed DAH is reported by bilateral funding source; but, in order to avoid double counting, the OECD-CRS and our other data sources report all ML DAH projects financed by core funding, by ML agency not funding source. Therefore, in order to identify which part of this ML DAH originates from the EU addition, we used ML core contribution data as report to the OECD-DAC at an aggregate level and then allocated these contributions to the health sector according to the proportions reported by MLs in the OECD-CRS.

Allocation of regional DAH

A large proportion of development assistance cannot be allocated to particular countries as it is provided at a global or regional level. There is also insufficient supporting information to allocate to countries both in the EU internal systems and in OECD-CRS project descriptions. We therefore have not allocated these projects to specific countries, but left them as regional unspecified. This underestimates the true development assistance flows reaching each partner country. We used the World Bank Classification based on Gross National Income level (Atlas method) to allocate countries to different income groups (low-, low middle-, upper middle- and high- income).

Country case studies

This study also included four country studies (Bangladesh, Mozambique, Egypt, Burkina Faso) to provide insight into development assistance flows at the country level. Specifically, the aim of the country visits was to: 1) complete the quantitative/qualitative assessment of expenditure (with a focus on local interpretation of global data); 2) further assess the broader national context (national priorities and relationship to EU priorities; indications of national expenditure trends; volume of aid from other donors; use of aid instruments and coordination mechanisms); and 3) identify examples of best/worst practices in aid reporting in a variety of different contexts.

The country case studies were not designed to replicate global mapping, but to try and identify key gaps from stakeholders and secondary data sources. Prior to the visits, an extensive document review was carried out of all sector policy, plans, expenditure frameworks and expenditure reviews (also supplied by the EU country delegations). Additional data was sought from National Health Accounting (NHA), Ministry of Finance expenditure reports, and public sector expenditure reviews. Where we could not get data on public expenditures to match we used the official government data. For other domestic expenditures we primarily relied on NHA data from local and if not available global reports. In addition, a web search was carried out to find project documents for all the large projects we had identified.

This was followed by country visits carried out between December 2011 and May 2012, each between 1 and 2 weeks, to conduct on interviews with key stakeholders. Key stakeholders were identified in close co-operation with the EU delegation, and based on a list provided by the study team; some respondents were referred by other stakeholders. Meetings were held with donor representatives, government officials (from sector and finance ministries) and representatives from NGOs (see detailed country reports for a list of key stakeholders interviewed). In total 91 stakeholders were interviewed. All freely agreed to be interviewed and their anonymity has been safeguarded.

The 1 to 3 hour semi-structured interviews were conducted using a topic guide. The interviews were conducted by interviewers with training and experience in health and education sector development and financing. Extensive notes were taken and many of the interviews were recorded. Where feasible, additional stakeholder focus group meeting were held in some countries, and key findings were reported back to stakeholders. In addition, to further contextualise and understand the response, a number of additional interviews were then conducted with stakeholders who were not available during the visit and, in some cases, with key experts in a particular field. Finally, further documentation that was obtained during the interviews was also reviewed.

Results

Global mapping

Between 2006 and 2009, the EU and its Member States disbursed around €20.4 billion in DAH in low- and middle-income countries (Fig. 1).

Fig. 1.

Global development assistance to health by type of source and year (millions of 2009 €).

This is in the context of collective EU development assistance disbursements of approximately €59.9 billion development assistance disbursed to the health sector globally. The proportional contribution of the collective EU to the health sector (34%) is substantially lower than in other areas, such as the education sector (66%) – primarily due to the relatively high levels of DAH provided by the United States, in particular for HIV programmes. EU sourced DAH commitments and disbursements to both the health and education sectors increased in real terms, in line with global aid trends. However, the rate of increase in EU DAH to developing countries was lower than the rate of increase by donors from outside the EU collectively. The EU Member States that made the largest DAH contributions were the UK, France, Germany, and the Netherlands. Other donors such as Spain, Sweden, Italy and Ireland also made sizable contributions. Figure 2 shows that the majority of collective EU DAH was disbursed through the EU Member States’ own bilateral development agencies and the EU's bilateral channel of assistance, although sizable amounts were also disbursed through ML agencies, in particular the Global Fund for AIDS, TB and Malaria and the World Bank (IDA). We also identified substantial funding through NGOs, particularly from France through Médecins Sans Frontières.

Fig. 2.

Collective EU DAH disbursement by source and type of channel of assistance 2006 to 2009 (millions of 2009 €).

Figure 3 presents the pattern of global DAH provided by NGO, bilateral and ML channels of assistance. It can be seen that most growth is observed through increased channelling through NGOs, although the role of multi-lateral agencies has slightly increased. It should also be noted that significant amounts of DAH reported here as bilateral development assistance are subsequently provided to ML agencies to implement development activities.

Fig. 3.

Global Development assistance to health by type of channel of assistance and year (millions of 2009 €).

Figure 4 shows how each of the top 20 collective EU recipient countries are funded by the different EU bilateral channels of assistance. Countries such as India, Egypt, Afghanistan and Indonesia benefit from particularly large contributions by specific donors. However, most of these top 20 countries appear to benefit from the interest of numerous EU Member State bilateral agencies; and in this respect can be seen as ‘donor darlings’, but also may suffer from highly fragmented DAH at the country level.

Fig. 4.

Top 20 recipient countries collective EU DAH disbursement by bilateral channel of assistance 2006 to 2009 (millions of 2009 €).

Conversely, some countries may also be considered ‘donor orphans’. Figure 5 shows the per capita disbursement of DAH over the entire period 2006 to 2009 for the 20 low- and lower middle-income countries which received the lowest per capita aid globally. Globally Iran, Columbia, Algeria, Belarus and China received the least per capita development assistance to their health sectors. Of the bottom 20 countries, the EU contributes a significant proportion of their limited DAH in Syria, Serbia and Burma.

Fig. 5.

Bottom 20 recipient countries (in terms of per capita DAH) DAH disbursement by type of channel of assistance (millions of 2009 €).

Most collective EU DAH was provided to low-income countries but this proportion varies considerably by donor country. The EU provides slightly around 50% of its total DAH to low income countries (and 85% to low- and low middle-income countries combined). This is in part driven by the relatively high levels of development assistance provided to countries such as Egypt and South Africa. It should be noted though that this overall figure does not take account of the distribution at the country level – and the recent thematic evaluation noted the good practice by the EU of geographically targeting specific areas in several countries in order to reach the poorest. Several EU Member States, Sweden, the Netherlands, Slovakia, Estonia, Denmark, Belgium and Ireland, provide over 70% of their DAH to low-income countries. Almost 30% of development assistance to health provided by the EU and its Member States goes to fragile states, as defined by the OECD.2

The EU provides the majority of its DAH to basic health care (60%). The second largest expenditure category is health policy (20%). Ten percent of its DAH is allocated to population and reproductive health, and another 10% to HIV/AIDS. A large proportion of collective EU (over 20%) and global DAH (over 30%) cannot be allocated to a specific sub-sector. If it is assumed that this unallocated DAH goes towards basic health services then there is less difference between the sub-sectoral allocations of DAH between the EU and other EU donors. We found some differences in the priorities of different EU Member States. The UK and France, for instance, allocate relatively high proportions to health policy, education and research, and Germany and the UK relatively high proportions to sexually transmitted diseases and HIV/AIDS.

Country case studies

We identified several additional sources of data on DAH available locally. In most countries, ‘on-budget’ development assistance had been captured in public expenditure reviews, although these were not available routinely. For the health sector, sporadic data was available from National Health Accounts (NHAs) (although more routinely in Burkina Faso and Egypt than in other countries). Some countries were also participating in National Aid Spending Assessments supported by UNAIDS. The degree of institutionalisation of different expenditure tracking varied widely by country, but for the most part NHAs were not institutionalised within any domestic governmental structure. However, most respondents were aware that a NHA had taken place and had used some of the data within it.

In addition, in Burkina Faso and Mozambique we found efforts to institutionalise development assistance reporting in government ministries. In Burkina Faso, there are several sources of information available. These include the ‘Direction Générale de la Coopération’ (General Directorate for External Cooperation) in the Treasury, which is responsible for the follow-up of the external aid and the elaboration of the annual report on cooperation for development. In addition, the ‘Direction Générale de l'Economie et de la Planification’ (General Directorate for Economy and Planning) (DGEP) in the Ministry of Finance is responsible for the follow up of Public Investment Programme (PIP) of the state as part of its role of monitoring government economic policies. Mozambique also has various mechanisms to track partner health expenditure. Since 2000, the Ministry of Health, with support from donors, has systematically gathered information on donor expenditure and pledges through surveys. This system is now called Inquérito de Fundos Externos (IFE) and is updated annually. Sophistication and comprehensiveness have improved in each update and the IFE is now a well-established monitoring tool, providing details including the type of modality used by donors, the on-budget status of each contribution, geographic information and the level of care distribution. The information collected in the IFE is used not only to produce sector analysis, such as the NHAs, but also for planning, since it is used for the Medium-term Expenditure Framework (MTEF), which forecasts future financial envelopes and expenditure.

An additional tracking system in Mozambique is ODAMOZ, an online-based tool managed by the Ministry of Planning and Development to collect data on partner contributions in all sectors. The process of data gathering is based on auto-reporting of partners and the website (www.odamoz.org.mz) supporting the database allows the production of customised reports aimed at boosting transparency and allowing managers and researchers to have access to this data. Unfortunately, not all donors report to ODAMOZ, undermining the effort and resulting in the ODAMOZ database not being complete. However, the remaining donors are reportedly considering joining this platform in the coming years.

Overall we found the amounts in our global mapping database to be broadly consistent with those found from local data sources, but there were several key omissions; particularly on funding channelled through NGOs. The other key omission was data from the Asian Development Bank (we were unable to obtain disbursement data by country from the Asian Development Bank globally) and the Islamic Development Bank. We also identified additional projects by a number of other donors who do not report to OECD-DAC, including China and Cuba – but we were able to obtain limited ad hoc information on expenditures from these donors at the country level. On the other hand, we also found that the global dataset was more comprehensive than local reporting mechanisms. Not all projects were captured at the local level, particularly contributions from smaller donors or those who may not have local offices, or whose aid flows from regional organisations.

However, although the broad amounts were consistent, local level data is considerably more comprehensive when describing the character of DAH flows. We identified a small number of additional straightforward coding errors beyond those already identified in the line by line review. We were also able to better identify DAH from different donors contributing towards a specific sector programme or pooled fund. Overall we found that global level DAH can appear to be more fragmented than it is, as often sector investments or spending within one project appears as multiple project lines. We also found the channel (of implementation) coding in the global dataset to be inaccurate for several projects in all countries, with most projects being coded under ‘other’, when it was often clear at the country level that it was going to a particular channel. The main concern expressed by respondents about OECD-DAC reporting at the country level was double counting funds channelled through intermediaries such as UNICEF. However, we were not able to verify the extent of this during our country visits.

Finally, it should also be noted that despite the existence of reports and systems, we also found a very mixed picture regarding the use of this country level data. Although the use of data from routine systems was more widespread than one off exercises such as NHA, we found many instances where donors and government (staff at the sectoral level) were not aware or using it. Where there was a dearth of local DAH data, surprisingly few of the respondents mentioned this to be a major impediment to sector co-ordination, planning and monitoring. However, where both the production and use of data had been institutionalised, particularly within sector review processes, there was much more awareness of the data by the donors and the government.

Discussion and conclusions

Our findings above show a complex picture of DAH flows – from source, to channel of assistance, to channel of implementation – that is hard to track at the global level, and rarely comprehensively and regularly tracked at the country level. While the majority of EU DAH is focused on low and lower middle income countries there also remains much disparity between countries; and further analysis is required to better understand whether these imbalances are fair and efficient. This data was presented to two meetings of lead health advisors from EU member states and can be used by donors and others to begin to identify possible overlap and areas for improved harmonisation.

However, while the reporting of EU Member States to OECD-DAC has much improved in terms of comprehensiveness in the past decade, we also find there is still room for improvement (12). Most EU donors compile their OECD-DAC extract from project management systems that may not neatly fit the OECD-DAC requirements, with various staff entering the original data. While staff may be given guidance on how to enter data, it is clear that there are many aspects that may be interpreted differently at the data entry level. This process needs to be quality controlled internally. For example the quality of reporting from the European Commission has substantially increased since 2006, following the establishment of review procedures and the systematic checking of the OECD-DAC submission.

Furthermore comprehensive sub-sector coding in health presents an insurmountable challenge unless it is possible to use a matrix approach (as in NHA) or enter multiple purpose codes. Without this it may be hard to identify where (if any) duplication of effort occurs. However, it is questionable whether efforts should continue at the global level in this regard – or focus on developing a new categorisation related to levels of the health system. Given the centrality of strengthening health systems to EU development policy in health, expending resources on separate sub-sector and disease expenditure tracking efforts is less important than tracking by different health system or sub-national levels. Health system functions could also be further clarified in line with recent analytical frameworks.

In addition, there are several areas that would assist or complement OECD-DAC. Possible supplementary exercises could include the following: a complete assessment of all EU Member States’ contributions as a source of development assistance through ML agencies; and development assistance flows from NGOs registered in EU Member States. The EU could consider establishing regular surveys to complement the OECD-DAC data in this regard.

Although much of the focus of development assistance tracking is at the global level, if harmonisation is to be achieved in practice, context is defining, and the real need and gap is therefore at the country level. There is considerable room for improvement in the country level tracking of development assistance to health and education. Given the large amounts of DAH funding overall, proportionally modest investments should be made in a number of countries, linked with sector programming to track DAH. Although these national efforts should be relevant to the country, where possible it makes sense to build on global efforts and encourage some compatibility of data across countries. A possible starting point would be to do trial projects in a few countries, based on experiences such as those in Mozambique for the health sector, facilitated by the EU, but with some attention to developing a system that has the potential to both draw from and feed into both OECD-DAC and NHA efforts, rather than starting with definitions and coding afresh. Unlike the OECD-DAC, however, national level efforts need to link into national planning timeframes and budgetary structures to be of use in national policy and planning – and if the two systems come into conflict, then the national requirement should always take priority.

Finally, few EU donor respondents in our country studies demonstrated knowledge of the issues in development assistance reporting. In addition, even where there was data available, few were aware of levels of government funding to the sector or how to assess areas such as sector and sub-sector expenditure. This may be a sign of over-stretched resources in donor offices, and the fact that those working at a sectoral level do not have a background in finance or economics. Therefore further support may be required at the country level to encourage all development partners to engage more fully in development assistance tracking efforts.

Conflict of interest and funding

This study was funded by the European Commission. The opinions expressed in this document represent the views of the authors, which are not necessarily shared by the European Commission or by the authorities of the countries concerned. The authors have not received any funding or benefits from industry or elsewhere to conduct this study.