Have a look at Dani Rodrik's blog where he discusses some opinions on the recent growth phenomenon of several African Economies, specifically linking it to an article by Ted Miguel. He links his discussion to Randomized (Impact) Evaluation, something which I believe is a very useful tool for development economists and should be used more for understanding what could work in SA specifically in terms of development policy.

As he emphasizes, what is becoming clearer and clearer is that:
1) finding out what DOESN'T work is very, very useful. This is something for which RE is particularly good (and which stops random politicians trying stupid policies that have been shown not to work)
2) what is often required for development is adherence to a set of ideas about how to attempt to implement policy rather than necessarily what the policy should be

Rodrik is somewhat sceptical about RE and has written an article about it. I will report back after exams on his academic article and on the Miguel Article. On Miguel he has done some really good work, and some really arb work. I saw a working paper of his recently, which was presented by a co-author from EUI at a UniSi seminar and I wasn't particularly impressed, but hey everyone has their ups and downs.