In this paper, I replicate the analysis in Acemoglu et al. (2001). These authors famously show that institutions are the primary determinant of differences in income between rich and poor countries, while geography has no effect on income, once in- situations are properly controlled for. However, if global income data are drawn from a bi-modal distribution, the effects of these explanatory variables may be obscured by the use of linear estimation models that implicitly assume these effects to be homogeneous across all countries. I use a range of analytical techniques, from simple sub-sample analysis, to sophisticated empirical estimation based on a finite mixture model approach in attempting to identify the effects on the AJR results of accounting for the bi-modality of the global income distribution. My analysis shows that the AJR results are sensitive to sample composition. In particular, it appears that institutions do not explain the variation in income across relatively poor countries or those in sub-Saharan Africa. I use Monte Carlo simulations to confirm that ignoring the bi- modality of global income could cause geography to appear unimportant. My findings could potentially reconcile apparently conflicting results from the existing literature on the role of geography and institutions in comparative development.


McDermott, T.K.J. July 2012. Geography and institutions: on the origins of comparative development. Under review.

Keep in touch with the Grantham Research Institute at LSE
Sign up to our newsletters and get the latest analysis, research, commentary and details of upcoming events.