William Easterly did not like Paul Collier’s The Bottom Billion (pdf):
The image of the trap is reinforced by Collier’s alarming statement that the bottom billion are falling further behind the rest of us. So is there a poverty trap—ie, the poorest countries are condemned to the worst growth? No, this is yet another statistical misunderstanding.
If you pick out who are in the poorest 1 billion today, naturally they would be disproportionately likely to be those that had the worst growth of incomes over the previous decades…
So if you want to test whether there is a poverty trap, you need to look at whether those who were poor at the beginning of any period you want to look at were more likely to have poor economic growth than the rest afterwards. The answer is no.
Easterly is rather harsh in his review, but ends with a great quip: “Economists should not be allowed to play games with statistics, much less with guns.”