Résumé
The generalised Lorenz criterion is widely used for making welfare comparisons within and across countries on the basis of their income distributions. Experimental studies have challenged this way of proceeding by showing that the principle of transfers, which underlies the generalised Lorenz criterion, does not meet with widespread agreement among the public that theorists would have expected. We propose to substitute the non-deprivation quasi-ordering introduced by Chakravarty (1997) for the generalised Lorenz criterion. This criterion is less demanding than the generalised Lorenz criterion as it builds on a weaker version of the principle of transfers and it is therefore more likely to be accepted by the public. We use income data from the Luxembourg Income Study for 17 countries in order to contrast the generalised Lorenz and the non-deprivation criteria. Although the non-deprivation quasi-ordering is less decisive than the generalised Lorenz criterion, it is shown that the former approximates the latter surprisingly well.
► We empirically investigate the degree of decisiveness of the non-deprivation quasi-orderings. ► These criteria are less demanding and more informative than the Lorenz criteria, but theoretically less conclusive. ► Statistical inference is performed using the intersection-union and union-intersection procedures. ► The non-deprivation criteria approximate the Lorenz criteria surprisingly well.