But if researchers and practitioners don’t understand each other, it is not because the language they speak is different, but rather because it is the same. An academic looks to the evidence as interpreted from field material to support opinions. The practitioners justify their actions by referencing the most up-to-date development policies – policies that are based on academic research. Academics and practitioners are two sides of the same development industry coin.

To understand the conundrum of evidence-based policies, we must start with the three main assumptions of the development industry: that development can be ‘engineered’ by aid, that such aid is actually used for development and that outside practitioners can implement development policies that work.

All three are debatable. To start with, there is no historical precedent for aid-generated development – as is demonstrated by East Asia’s massive and primarily self-propelled surge forward. Nor is there solid evidence that aid is actually used for development. Many recipient governments, especially in Africa, have no coherent development policies or lack the capacity to implement them. Aid is thus primarily deployed as budget support and serves to keep administrations running. Finally, it is simply unrealistic to expect that outside practitioners can make a crucial difference.

One topical example is decentralisation. The notion that decentralisation might have a positive effect on development has dual origins. One is that the state, which centralises and controls resources, including aid, has failed to provide the local services and infrastructure that are needed for development. The other is that transferring resources directly where they are needed will help make the local people more focused on the development aims they want achieved. However, the evidence for this was found in settings where development had already taken place, such as Europe or Asia, and from some areas, like Kerala in India, where there has long been a strong and dedicated provincial administration. We simply cannot assume that decentralisation will be as effective elsewhere.

The flaw is in seeing causality where there is merely correlation. Where the state is institutionalised, and the administration relatively adept, the devolution of revenues and services can, in certain circumstances, help to combat poverty and support development. What the global evidence shows is that the more efficient the state, the more likely it is that decentralisation will contribute to achieving these objectives. However, to see decentralisation as the means to improving service delivery, increasing accountability and, especially, engendering development is to assume a causality for which there is scant, if any, evidence. Policies that are not evidence based run the risk of failure, thus further discrediting the usefulness of aid.

So what can be done? Three points need attention. The first is aid policy. More aid does not automatically mean more development. It may be necessary to give aid, especially in emergencies or conflict resolution, but, pace Jeffrey Sachs, there is no compelling proof that more aid in itself will spur development. The second point is that analysts must pay attention to the evidence, particularly about state institutionalisation, rather than listen to those who pay the consultants. A first step would be to study the historical record: why, for instance, has East Asia developed whereas Africa has not? Here, Collier’s The Bottom Billion is simplistic and naive. Finally, Western practitioners must begin to study the evidence for themselves and thus acquire the capacity to challenge the assumptions on which policymakers make decisions.