Good approximations to maximin allocations (Q797458)

The Rawlsian maximin welfare function differs from conventional welfare functions in two respects. Firstly it places a weight of one on the worst off individual in society. Secondly, as a direct consequence of this, the welfare function ignores almost everyone in society, i.e., it is possible to increase almost everyone's income and social welfare will not change. This note considers the implications of rejecting the first but accepting the second property. In a society where each individual has a differentiable, concave, bounded utility function and an endowment of income we postulate a welfare function that puts a weight of \(\mu>0\) on the worst off individual and a weight of (1-\(\mu)\) on a specified group of individuals. We show that for all \(\epsilon>0\) (no matter how small) in almost all of the economies of the type we have outlined, the minimum level of utility for any \(\mu\) is less than \(\epsilon\) below the maximin level of utility. Thus in the type of economy we are discussing the fact that almost everyone receives no weight in the welfare function is the crucial element in determining the minimum utility level rather than the weight the worst off person receives.