In the May 2007 issue of the American Economic Review, Bill Easterly argues that one reason why the overall package of development assistance offered by the World Bank and the IMF has not been very successful is that they (we?) mistakenly believe that they know what actions are needed to achieve development (ungated link here). First it was Investment that was the key to growth, then Human Capital, then Openness, then Institutions. As Easterly puts it: "Development economists have long known the answers on how to achieve development. The only problem is the answers change over time".
Meanwhile, in the September 2007 issue of the Journal of Development Economics, Robin and I provide empirical confirmation of this point by showing that while countries have become much more homogeneous in their policies (policy variables are converging), output paradoxically continues to diverge (ungated link here).
Specifically we show that Investment, Government Spending and Openness to Trade, and several measures of institutional quality are all converging in our sample of 90 countries from 1960-1999 while per capita incomes continue to diverge. A la Easterly, we interpret this as showing that poor countries have on average followed the development advice of the Bretton Woods Institutions (Bank and Fund), but have not gotten the promised payoff.
Going beyond this point, we also claim that our results show the neo-classical growth model to be totally inadequate to explain the data.