The Mirage of Convergence: Why Poor Countries May Only Seem to Be Closing the Income Gap

Results of numerous cross-country growth regressions have been found to be sensitive to specification, time period or sample coverage. Several authors have observed that results may depend on the source and data collection methods for right-hand side variables. In this paper we suggest that a more fundamental problem may exist with respect to the growth rates used in the majority of studies. Differences in measured growth rates are severe across widely-used sources. More critically, these differences are correlated with countries’ level of development. As an illustration, we show that the results of two recent studies depend critically on which data set is used to derive the growth measure.