Aid policy and evidence: is conditionality well-based?
Tony Killick, Senior Research Associate - ODI
Simon Maxwell, Director - ODI
1. Tony Killick made a presentation based on his recent paper in Development Policy Review . His main, overarching point was about the use of evidence in aid policy. The article discussed three cases in which he felt evidence was not being used correctly. He would focus on one of these, the case of conditionality.
2. The evidence indicated strongly that conditionality was largely ineffective, and indeed the current development discourse reflected this. In practice, conditionality had not decreased and might well have increased. This could lead to a large-scale waste of public resources.
3. The evidence that conditionality did not work was well established. Tony Killick's own earlier work had illustrated the high frequency of breakdown of IMF programmes and the poor correlation between the quality of aid countries policy and support by World Bank structural adjustment programmes. Structural adjustment, he argued, had been wholly ineffective in achieving either an improvement in policy or a high average standard.
4. It was important to understand why compliance with conditionality was poor. Part of the reason was that local politics tended to dominate the outcome, but it was also important that the incentives on staff in the international financial institutions were not such as to encourage them to pursue non-compliance. In fact, the threat of aid withdrawal was not usually carried out.
5. Since the early 1990s, the conventional wisdom had been to reduce conditionality. However, the number of conditions in World Bank and IMF programmes had only fallen very slowly, and those that were legally binding had fallen least of all. Furthermore, a new kind of conditionality had been introduced through Policy Reduction Strategy Papers.
6. Tony Killick concluded that it was important to find alternative ways of doing business.
7. The discussion spent some time analysing different kinds of conditionality. Some participants commented that there was now a high degree of consensus between donors and e.g. African governments on the way forward, and that therefore most conditionality was now consensual. Furthermore, there was a difference between conditionality on substance and conditionality on process, with the latter being less damaging and difficult. Tony Killick was somewhat sceptical of these arguments. He argued that conditionality related principally to policies governments would either not implement at all or not implement quickly enough. He described these as "hardcore" conditions, and suggested that the earlier conclusions still applied. He also thought it unlikely that the new consensus resulted in any way from conditionality. The World Bank itself had found in countries like Mexico and South Africa that patient dialogue and influence was a better way to change the mind of government. He thought partnership might be a better way forward than conditionality.
February 23, 2004
In 'Politics, evidence and the new aid agenda', published in the Development Policy Review, January 2004, Tony Killick questions the use of conditionality as a means for achieving policy change in aid-recipient countries:
"On the one hand, much research - and quite a lot of Bank (but not Fund) rhetoric - emphasises the limitations of conditionality as an instrument for change. On the other, it is arguable that, at least within low-income indebted countries, governments find themselves expected to conform to an even wider array of policy stipulations than in the apparent heyday of conditionality in the earlier 1990s. However, the issues here are complex and the evidential basis less satisfactory. They are also important, because if misplaced reliance is placed on an instrument which actually fails to deliver the safeguards it appears to offer this can lead to the mis-application of large amounts of public money".This event pressed on the use of evidence in aid policy. It discussed cases in which evidence was and was not being used correctly.