When the great and the good gather at the Millennium Development Goals Review Summit in New York in September, it will be interesting to see how prominent general budget support (GBS) is on the agenda. Back in May, a paper on ‘Budget Support and MDG performance', by Jonathan Beynon et. al. was showcased at ODI, showing that countries that have received large amounts of GBS have also performed better against selected MDG indicators than those who have received little or no GBS.

There is growing interest in the results-based aid agenda and the achievement of the MDGs. The EC has already made the jump to tie GBS provision to the MDGs and is pioneering a results-based approach to aid in the form of the MDG-contract , whereby a MDG-tranche of at least 15% of the total budget support commitment will be used specifically to reward performance against MDG outcomes. This focus on aid outcomes was not far from the mind of Andrew Mitchell, UK Secretary of State for Overseas Development, when he announced the introduction of the new independent watchdog , saying: ‘We need a fundamental change of direction – we need to focus on results and outcomes, not just inputs'.

However the pressing research challenge is that the evidence to support the focus on linking aid modalities to certain types of results has been limited to date. Some proponents of the new MDG-contract will be glad to see that aid is now being tied to poverty and social welfare indicators, considered to be the ultimate goal of development initiatives, on the assumption that recipient countries will be further incentivised to achieve the MDGs.

Yet one sticking point is that analysis and evaluations of GBS have often stopped short of attempting to assess the effects of GBS on poverty and social welfare indicators, instead only tackling the more immediate effects of GBS on areas such as country systems, government capacity and service delivery. Those that have attempted this have often acknowledged the difficulties of tracking the distinct effects of GBS on poverty . This is because explaining the relationship between aid instruments and poverty outcomes requires the use of complex intervention frameworks where explicit judgements need to be made about attribution. Even when useful frameworks are developed, testing them is very difficult in an environment where quality data is notoriously difficult to come by. This challenge was made clear in ODI's recent work on cross-country evaluation of Sector Budget Support and a recent Options Appraisal of aid modalities in Malawi.

The new OECD-DAC GBS evaluations will attempt to analyse and unpack the association between GBS and poverty outcomes, as outlined in its methodology , but it is unlikely that the results of the evaluations will be available for a couple of years. In the meantime, economic conditions will require tough decisions about aid and it is likely that those decisions will be guided by performance. A better understanding of the complex relationship between aid modalities and their effects on outcomes is needed now to inform such decisions. This is why the paper by Beynon et. al., and ODI's public meeting and the continuing debate on this issue are so important. There is need for much further work on the question of causality, but we are seeing the welcome start of a much needed research agenda.