Global Divergence in Growth Regressions

This paper examines the possibility that a country might follow the global technology frontier partially instead of fully. We develop an empirical test of this possibility, which measures by how much a country follows the frontier. The test shows that indeed most countries do not follow the frontier fully. This means that they diverge away from the frontier. Hence, this paper suggests that convergence should be measured not only by the traditional convergence parameter, which actually measures convergence to the long-run path, but also by the parameter that measures by how much this long-run path follows the frontier. Finally, our approach enables us to disentangle the effects of various explanatory variables to separate effects on the rate of departure from the frontier and on the level distance from the frontier.