I just read Andy Sumner's latest IDS working paper "Global poverty and the new bottom billion: Three-quarters of the World’s poor live in middle-income countries". It is quite a straightforward but carefully done mapping of global poverty.
What I find most interesting is not that 3/4 of the world's poor live in middle income countries (mainly due to India and China crossing middle income threshold, currently set at $995) but the results that are net of China and India. (Note that the data exclude the 105 million people who live in Iraq, Afghanistan and Sudan as these countries have no poverty data).
Once you take India and China out, we can see that low income fragile countries share of global poverty falls from 13 to 12%; low income non fragile rises from 12 to 16%, middle income fragile increases from 1 to 11% and middle income non fragile increases from 6 to 11%. Overall low income increases from 25 to 28%, middle income from 7 to 22%, fragile from 14 to 23% and non fragile from 18 to 27%. The slide here makes it clearer.
So the imperative to target aid to the lowest income countries is in tension with the desire to use aid to address fragility. It is the "middle income fragile" category for which the share of global poverty is growing the fastest (countries have the means but not the ability to combat poverty) and this is where new innovations for using aid to leverage development will be found to make aid more effective.