ABSTRACT
The evidence on sigma-convergence in income indicated by (a) coefficient of variation (CV) and (b) SD of logarithms (SDLOG) is considered for a large cross-country sample covering the period 1960–2010. Three main points are noted. First, the two measures yield qualitatively similar scenarios, and both indicate sigma-divergence in income over the period. Second, however, they do show large differences in the rate of change in income inequality, and SDLOG indicates divergence at a much higher rate than CV. It seems likely that SDLOG would indicate greater divergence, or weaker convergence, than CV in many cases. Third, therefore, researchers are urged not to rely too heavily on one or the other measure for an inference on sigma-convergence, and it seems appropriate to consider both for drawing reasonable conclusions on convergence in income and many other variables studied by scholars.
Acknowledgement
Insightful comments from a perceptive reviewer are gratefully acknowledged. The usual disclaimer applies.
Disclosure statement
No potential conflict of interest was reported by the author.
Notes
1 As Milanovic (Citation2005, 4) and Ram (Citation2017) show, while unweighted measures of cross-country income inequality indicate divergence, weighted income-inequality indexes suggest convergence.
2 Additional details about the illustrative weights for CV and SDLOG calculated for the year 1980 from the data used in this study are available from the author.