Reviewing Paul Collier’s The Bottom Billion, William Easterly writes,

Collier comes perilously close to another statistical fallacy known as selection bias. He chose the countries that belong to his Bottom Billion on the basis of their poverty today, and then points out that they also have had very poor growth during the preceding four decades–as has been the case, for example, with Angola, Haiti, Liberia, Sierra Leone, Somalia, and Zaire/ Congo. The implication is that the poorest nations will in the future continue to have very low economic growth.

But there is actually no evidence that the Bottom Billion at the beginning of a forty-year period will have worse economic growth over the subsequent forty years than rich nations. What Collier did was select which countries are poor at the end of the period, so of course they will also be the ones that previously have had a long string of very low growth rates–if they had previously had high economic growth, then they wouldn’t be poor at the end.

Easterly spends much of his review expressing skepticism about the use of outside military intervention to relieve severe cases of poverty caused by civil war. I should emphasize that while Collier does make some recommendations along these lines, they do not constitute the core of his book. Read the entire review, but also read the entire book.

Thanks to Mark Thoma for tracking down the ungated version of Easterly’s review.