Abstract
In this study we explore the evolution and origin of international welfare inequality in a sample of 39 countries between 1970 and 1998. To achieve this aim, different theoretical results taken from the literature on personal income distribution have been applied to an additive decomposition of Sen's welfare index. In relation to the evolution of international disparities in welfare, the conclusions obtained vary depending on the inclusion in the analysis of population shares. In addition, cross-country productivity differences emerge as playing the major role in accounting for observed patterns of welfare inequality.
Notes
For an example, see Theil (Citation1996), Duro and Esteban (Citation1998) or Firebaugh (Citation1999).
For a detailed analysis of this issue, see Grün and Klasen (Citation2002).
This index was used, for example, by Ram (Citation1992) or Duro (Citation2001).
It is worth pointing out that the variance of the log can be expressed as the square of the standard deviation of the log, which is widely used in the growth literature to measure the concept of sigma convergence. In connection with this, see for example Barro and Sala-i-Martin (Citation1992).
This is what Shorrocks (Citation1982) terms the natural decomposition of the variance.
These countries, a complete list of which is included in the appendix, represented roughly 70% of the world population in 1998. For cases in which the Gini index for a given year is unknown, we have used the temporally nearest available value. With respect to this issue, it is worth mentioning that the results obtained by Li et al. (Citation1998) indicate that the degree of dispersion in personal income distribution in the various countries tends to be fairly stable over time.
We have estimated the degree of statistical association between SW and its various components. In this respect, the mean values over the period of the Spearman coefficient of correlation between the welfare index and productivity, per capita employment and inequality in personal income distribution are 0.956, 0.419 and 0.367, respectively, which confirms the results found earlier.