ABSTRACT
Although the properties meanings and limits of the Lorenz dominance criterion are now well understood in the univariate case (see, for example, Atkinson, 1970; Kolm, 1969; Sen, 1973) by the theorists and by the practitioners (see for instance, the diverse entries in the Handbook of Income Distribution (Atkinson and Bourguignon, 2000)), it is not the case for its various extensions to a multiattribute context. However, the pressure is growing for such extensions, thanks to the widespread accessibility of data bases where the description of individual units is plural. Besides, a consensus is emerging among many scientists, particularly development economists, about the multidimensional aspect of individual wellbeing which cannot be reduced to a unique monetary dimension. The economists’ awareness of this multiplicity traces back to the ethical analysis of Rawls (1971) and Sen (1985, 1987, 1992). The lack of health attributes in any study on poverty now appears as a major drawback. Education appears as another important dimension which is even more complex to apprehend owing to its impact on income streams in dynamics. It is important to stress that many additional aspects of well-being may not be compared on a true cardinal basis and often a ranking of individuals with respect to these non-pecuniary aspects is the only material to work with. Despite the fact that we are aware of this paucity of information, it is assumed in the following that all attributes are given a cardinal meaning. A solid defense (at least for a review) of such a line comes from the fact that very little is known in a true ordinal setting (see however Allison and Foster (1999)).
