A consistent multidimensional Pigou-Dalton transfer principle

The Pigou-Dalton principle demands that a regressive transfer decreases social welfare. In the unidimensional setting this principle is consistent, because regressivity in terms of attribute amounts and regressivity in terms of individual well-being coincide in the case of a single attribute. In the multidimensional setting, however, the relationship between the various attributes and well-being is complex. To formulate a multidimensional Pigou-Dalton transfer principle, a concept of wellbeing must therefore first be defined. We propose a version of the Pigou-Dalton principle that defines regressivity in terms of the individual well-being ranking that underlies the social ranking on which the principle is imposed. This well-being ranking (of attribute bundles) is induced from the social ranking over distributions in which all individuals have the same attribute bundle. It is shown that this new principle—the consistent Pigou-Dalton principle—imposes a quasi-linear structure on the well-being ranking. We discuss the implications of this result within the literature on multidimensional inequality measurement and within the literature on needs.