Here's a crucial exchange from the Committee of Public Accounts' grilling of DfID staff about aid to Malawi:

Chairman: We know that if you put enough money into any scheme you will achieve something, but we are a value for money committee, we are looking at efficiency...

Dr Shafik: ...I think the NAO Report does also say that the DFID programme in Malawi has contributed clearly to poverty reduction...

Chairman: I am not arguing about that. That was precisely why I said what I said. If you spend enough money you are going to achieve something, but what I want to know is how can we as a value for money committee be assured that you are achieving value for money when clearly you are lacking in data about how effective your programmes have been in terms of value for money?

But obtaining better data on aid effectiveness is two-fold:

The easy part is to demand more evaluations and to demand that they are of higher quality.

The difficult part is designing interventions which are evaluable in the first place. That is, funding projects which can be subjected to a clear, fair and unambiguous test of whether they have succeeded or failed.

How much will aid have to change to become evaluable?