Tuesday, November 27, 2007
A Review of Collier's The Bottom Billion
William Easterly reviews the book in The Lancet.
Collier does mention that causality between actions and outcomes usually fl ows in both directions, but never explains how he then establishes the causal eff ect of actions on outcomes that form the basis for his remarkably confi dent statements.
There are also lots of other tricky problems, like controlling for other factors that might be triggering both foreign actions and local outcomes—say, good or bad weather, or collapsing or soaring commodity prices. The apparent eff ect of an aid action may just be a stand-in for some other factor—for example, an economic recovery after a drought may occur at the same time as the usually tardy drought-related foreign aid arrives, generating a spurious correlation between aid and economic recovery. Controlling for third factors is both crucial and diffi cult, yet again Collier has little to enlighten the reader about just how he arrived at his assertions on which actions cause which outcomes.
Others have noticed these problems. Much of Collier’s civil war research was done when he was at the World Bank. A later evaluation of World Bank research by a blue-ribbon panel of economists, led by Angus Deaton of Princeton, singled out Collier’s civil war research for criticism on these same grounds. Deaton’s panel concluded that the “analyses in these studies cannot be used to support the conclusions that they ostensibly reach”. Massachusetts Institute of Technology economist Daron Acemoglu was part of the panel and wrote the evaluation of the civil war work, saying “the correlations that are interpreted as causal eff ects are really no more than correlations”.
The problem of correlation versus causation pervades the book. “The bottom billion” are—no surprise—poor. Again, Collier presumes poverty causes war (what he calls a “conflict trap”). Poverty and war do seem to go together, but Collier fails to offer convincing evidence that a given amount of poverty relief (however that would be accomplished) would cause reduced war. And the threat of spurious correlation is still a problem, as poverty and civil war may go together only because they are both symptoms of deeper problems, like Africa’s weak states, ethnic antagonisms, and the legacy of the slave trade and colonial exploitation. His shaky analysis leads to real world advice (like foreign military intervention to break the “conflict trap”) that could be tragically wrong.
Posted by Robin Varghese at 05:11 PM | Permalink