CAFOD have been looking at the figures too – and we actually came up with a few different narratives on the trends. As Andy Sumner says, you can always play with statistics to come out with alternative story-lines, so it’s no surprise really. Our raw data is here.
Both CAFOD and IDS did a ‘before and after’ analysis of UK bilateral spending, and we each struggled with the question of what data to use, in order to see the picture ‘before’ the BAR. In the end, CAFOD decided to use the latest spending figures that DFID had on their website (UK net bilateral spend for 2009 from the 2010 Statistics for International Development – aka ‘SID’), rather than just the BAR data (which gives bilateral spending for 2010/11 but only for countries that will continue to receive some funding in future years).
For us, the incompleteness of the BAR data – excluding countries such as China, Niger, Angola, Gambia, Lesotho – was a more serious concern than the issue of whether the 2009 data was comparable. It’s a judgement call. IDS decided to go the other way.
If you use only the BAR data, as the IDS team did, it looks like more bilateral aid spending is going where more poor people live (the £ per poor person is rising in Africa and Asia). However, if you use the SID data, it looks like the opposite is happening. Our analysis suggested that the correlation between aid spending and the number of people living under $1.25 per day actually fell from 0.63 to 0.45 (p<0.05). Basically, DFID funding is being cut to several countries that have very high poor populations – China and Indonesia most notably. If you use the BAR data, which excludes China, you create a blind spot to this.
We were also measuring slightly different things. IDS were focusing on the aid spend per poor person, whilst we were interested in the correlation between the number of poor people living in different countries, and the amount of aid those countries received. It’s also a judgement call to say which measure matters most.
CAFOD also came out with different findings on whether more aid is going to more fragile states, and whether it is going to more corrupt countries. We used a different measure of fragility (IDS used the fund for peace index, whilst we joined with DFID in using CIFP), which probably explains some of the contrast. We found that the correlation between fragility and aid allocations is only fractionally stronger after the BAR – rising from 0.34 to 0.35 (p<0.05). The stability of the correlations is because DFID is not only increasing spending to a few fragile states, but also withdrawing from some others. Iraq, Angola and Niger score pretty highly in the fragility stakes – and the fact that DFID is pulling out here provides something of a counterbalance to the increases in Pakistan, Nigeria and Somalia. If you don’t use data that includes these countries, you don’t see the pattern.
The same thing happens when you look at corruption. Using the SIG data, we found that the weak negative correlation between corrupt countries and aid allocations grew very marginally after the BAR (from -0.10 to -0.12, p<0.05). DFID is giving money to many countries with high levels of corruption – but it has also stopped giving money to many countries with high levels of corruption. The overall risk doesn’t look very different. The narrative here would be: “It’s always challenge to spend aid where there is a lot of corruption, but there hasn’t been a dramatic rise in the risks for DFID”. Bit boring, really.
The thing that everyone seemed to be worried about before the BAR – the risk that the fragile states focus would become a cover for allowing UK security interests to dominate aid allocations – seems to have dropped off the BAR blogosphere altogether. This is probably because the answer is a bit boring. Basically, it doesn’t look like there’s an overall bias towards security interests in the allocations. Nobody I’ve spoken to seems to think otherwise. Shame that we don’t bother writing about good news, just there’s an element of damp squib about it.
So can we juice things up by throwing in some different questions, and looking at how much aid is going to “not Free” societies? It’s certainly got some timely tie-ins, with the Libya crisis and the middle-east revolution. The methodological issues really are pretty serious though. My inner anthropologist has held her nose to the point of passing out.
The decision on which data-set to use means that CAFOD and IDS have come out with slightly different narratives on what the BAR results mean, but there’s no doubt that we’d agree on one thing in particular: It would have been much better if DFID had published its BAR results as a spreadsheet, correctly added up, with all the data online when it mattered, and in a format that was comparable with previous years. Given all the talk of commitment to meaningful transparency, it was a bit surprising that this didn’t happen.
Lets hope that when the country-operational plans come out, we don’t have similar data-problems.