The EU Code of Conduct on Complementarity and Division of Labour in Development Policy, approved by the Council on 15 May 2007, is potentially revolutionary, with significant implications for the future of British and other bilateral aid.

The Code of Conduct is based on background work by the German Development Institute, and has also been the subject of a Commission Communication to the Council and the Parliament.

The final document lays down 11 principles. These are

1. Concentrate on a limited number of sectors in-country, effectively a maximum three per donor per country, plus budget support;
2. Redeploy out of other sectors;
3. A ‘lead donor’ arrangement, whereby one EU donor leads in each sector;
4. Delegated cooperation/partnership, in which donors engage in sectors over and above their chosen three through another donor, to whom they delegate authority for policy dialogue and administration of funds;
5. Adequate donor support, but limiting the number of donors in any sector to a maximum of 3-5;
6. Replicating the above at regional level and with regional institutions;
7. Establishing priority countries for each donor, to avoid spreading resources too thinly;
8. Addressing the orphans gap;
9. Analyse and expand areas of strength as between donors, in order to play to comparative advantage;
10. Pursue progress on other dimensions of complementarity; and
11. Deepen the reforms, by providing the right incentives and sufficient decentralised staffing.

The Code of Conduct is carefully written to emphasise a gradual and pragmatic approach. The Council stated that the code was ‘voluntary’. Nevertheless, the Council also said that the Code should be ‘applied immediately and progressively by the Member States and the Commission, building on existing systems’.

Of course, the general principle of better coordination is exemplary, and entirely consistent with the principles of the Paris Declaration on Aid Effectiveness, which emphasizes harmonization as the way forward for donors. Indeed, para 35 of the Declarations says that donors will ‘make full use of their respective comparative advantage at sector or country level by delegating, where appropriate, authority to lead donors for the execution of programmes, activities and tasks’.

The need is also obvious. Our own work at ODI on aid architecture has illustrated the costs of donor proliferation. A recent paper on architecture, prepared by the World Bank as a contribution to discussion about IDA 15, provides further evidence of ‘proliferation’ (= lots of donors) and ‘fragmentation’ (= lots of activities). For example:

1. There are at least 56 bilateral donors and over 230 international organizations, funds and programmes.
2. The average number of donors per country has nearly tripled over 50 years, from 12 in the 1960s to 33 2001-5. The health sector is especially problematic. 
3. There are 60,000 aid projects. The average size is only $1.5m.
4. Tanzania has more than 700 projects, managed by 56 parallel implementation units. In 2005, there were 541 missions of which only 17% involved more than one donor.

The challenge is perhaps particularly acute in Europe, with many new donors entering the system as a result of enlargement, and with European countries scheduled to provide 80% of the increase in aid promised at Gleneagles. There are other challenges also, including the large number of new donors outside Europe and the proliferation of special purpose ‘vertical funds’, like the Global Fund for HIV/AIDS, TB and Malaria. Simplification would certainly reduce transactions costs.

All this being said, the implications of the Code of Conduct are radical. In particular, all donors, including presumably large ones like the UK, should restrict themselves to a maximum of three sectors per country and either redeploy out of other sectors or work as a silent partner, allowing another EU donor to take the lead. A further constraint is that there should be only 3-5 donors per sector. The Code of Conduct does not say whether this means all donors or only EU donors, but assuming the more generous interpretation, that the constraint is the number of EU donors, that is still likely to limit engagement in popular sectors and countries: not everyone can choose health in Tanzania as one of their top three priorities.

In order to assess the long-term implications, we would need to know how many sectors each EU donor supports in each country. That’s not so easy. A prior question is what is meant by the term ‘sector’, which could be wide or narrow. The narrower the definition, the easier it becomes to meet the requirements of the Code. Thus, if a ‘sector’ could be defined as ‘health programmes targeted at the sexual health of adolescent girls in Shinyanga in Tanzania’, with similar programmes elsewhere making up different sectors, then there could be no limit to donor involvement in the health sector in Tanzania. On the other hand, ‘health in Tanzania’ is much more restricted.

The Code of Conduct is not precise on this issue, but does address the problem, It says:

‘The appreciation of what constitutes a sector, being intuitive or informed, should be done in a flexible manner, at partner country level and match the definition of the partner country, that should have identified the sector as a priority in its poverty reduction strategy or equivalent. In agreement with the partner country, the partitioning of sectors should be avoided as much as possible.’

That seems to rule out sub-dividing to the point of concentrating on adolescent girls in Shinyanga. It still leaves the question somewhat open, however, and variable by country. The Code does not say, as it might have done, ‘use international definitions of sectors, as defined by the DAC’.

Data to test the implications is presumably available on the DAC data base. I’m afraid I haven’t done the analysis. Someone should. For the EU, it would also be possible to check the EU donor atlas, though the published version does not break down aid by donor, sector and recipient all at once.

For the UK, there is information in British Aid Statistics and in the Departmental Report, the latest edition of which was published only this month. BAS does not seem to provide a breakdown by recipient and sector, and nor does the statistical section of the Departmental Report. However, Chapter 2 of the Report does illustrate DFID activities country by country and MDG by MDG. It is immediately obvious that in most of its high priority countries, DFID is engaged in many sectors. This is despite the move to budget and sector support, which accounted for 28% of bilateral aid overall in 2006-7 and over 80% in Ethiopia, Mozambique and Tanzania (See table 5.2 of the Departmental Report). In Ethiopia, for example, DFID contributed £52m to budget support for a programme to Protect Basic Services, and in addition, among other activities listed, provided

• £17m to a programme of safety nets;
• £2m for humanitarian aid;
• £1.5m for teacher training; and
• £1m for a programme on contraception.

In addition, DFID made preparations for a £100m programme on water supply and sanitation.

What this suggests is that greater concentration is likely to demand difficult choices. There are plans for specific exercises on the division of labour. The GDI paper reports on the first of these, in Zambia, with the results below. It concludes that:

‘The number of donors per sector was barely reduced. Even if the questionnaire in Zambia did not explicitly ask for a reduction of sectors and focussed on the interest of donors to take a lead function in a maximum of three sectors, there was a clear understanding that the number of donors should be reduced. In addition, there is a large number of lead donors in Zambia. This is partly due to sub-sector leads and the amount of co-ordination work resulting from the remaining large number of donors. Over time, the number of lead donors per sector could be reduced if trust in the lead donor concept is established, the number of donors per sector is reduced further and regular peer reviews of lead donor performance are established. It seems fair to conclude that the willingness of donors to withdraw from sectors they are currently engaged in is limited.’

Another study is close to completion in Uganda.

Although donors may be unwilling to withdraw, there are examples. For example, the Commission Communication to the Council on this topic, cited earlier, has this example:

‘In South Africa, the UK Department for International Development (DFID) has
delegated the implementation of its support to land reform to Belgium.’

It is interesting that ‘land reform’, rather than ‘rural development’ or ‘agriculture’ should be considered a sector: this is close to the ‘sexual health of adolescent girls in Shinyanga’ case. In addition, it would be interesting to know about the strength in depth of Belgian expertise on land reform. In a parallel case, anecdotal evidence is that staff turnover can be a real problem. In one country, DFID is said to have ‘delegated to the EC in Education, and the EC adviser promptly left and was not replaced’.

What do I think about all this? I am all in favour of simplification, but frankly, my long-standing and published view is that the best way to achieve that is by channeling a greater proportion of funds through multilateral organizations (see also this paper on the future of bilateral aid). The slogan should be ‘Don’t just harmonise, multilateralise’. The prospect of 56 bilateral donors and over 230 international organizations, funds and programmes, all manoeuvring to see which of them will be allowed to operate in specific sectors in particular countries, and which of them will lead on what, just seems to me like a recipe for bureaucratic gridlock and also a significant threat to quality. If aid does increase as we expect it to, then these problems can only become worse. Surely, the top priority for donors in 2007 should be to make sure the multilateral development banks, the UN and the European Commission are properly equipped to take the lion’s share of new resources? See, for example, this piece on future EU funding and also our work on the UN.