The logic of inequality extraction : an application to Gini and top incomes data

As embodied in the concept of "inequality extraction" (Branko Milanovic), it is not possible to increase inequality (especially income inequality) in a society sustainably to levels beyond what is actually socially acceptable (and even less to levels endangering physical subsistence of parts of the population). Consequently, in international and intertemporal comparisons of inequality levels on the country (or the regional) level and for poorer societies in general, adaptations for the levels of inequality, which are actually feasible in physical and social terms, are necessary to arrive at meaningful conclusions. We extend this concept to cover also top-incomes ratios, which allows a broader database for measurement and analysis of crosscountry historical inequality. A first look at the data uncovers different patterns and dynamics of inequality extraction when compared to untransformed inequality measures, sometimes to the point of trend reversal, which holds for extraction ratios based on Gini coefficients as well as top income shares. In the discussion, we also outline some directions of further research, especially about explanations of extraction patterns and dynamics, also with an application to global inequality data and in combination with institutional arrangements.