This paper was prepared for OED conference on Evaluating Development Effectiveness, July  2003.

In recent years, the pendulum of professional opinion about effective aid modalities has swung away from an original concentration on project-based assistance in favour of more programmatic forms, most notably budget support and the associated modality of debt relief. Although some donors remain wedded to the project mode, there is little doubt about the direction of trend. Thus, a recent British policy statement on the subject (DfID, 2000:93):

...there needs to be a real improvement in the way that assistance is delivered. That means reducing support for stand-alone projects, and increasing support for sector-wide reforms. Where governments have a strong commitment to poverty reduction and strong policies in place, it means moving towards providing financial support directly to recipient government budgets using their own systems.

Since the end of the 1970s the World Bank has, of course, been providing programmatic ‘policy-based’ adjustment credits but, in low-income countries, the more recent opening of its Poverty Reduction Support Credit (PRSC) window has added a further dimension, not least because -- to its great credit -- the Bank is beginning to go to some lengths to harmonise the terms of its PRSCs with those of budget support baskets put together by bilateral and other donors. The introduction of the Enhanced HIPC debt relief scheme in 1999 added a further string to the programme aid bow.

It is hardly contentious to suggest that this shift represents a challenge to evaluators. New issues arise and new needs for evaluation are created. One ‘big’ issue is the extent to which the shift from project to programme support is soundly based on evidence about the comparative effectiveness of different modalities: have past evaluations and research provided the evidence needed for informed decisions about modalities? Another issue, however, is the extent to which those who make decisions about modalities are actually driven by evidential factors: do they listen to evaluators?

Perhaps to some extent they are unable to because of time-lags. The contemporary aid environment is a rather fast-changing one whereas evaluation is retrospective, occasional and reflects the aid choices of the past rather than the fascinations of today. Finally, the switch to programmatic support raises questions about the level at which evaluation should be conducted and who should undertake it.