In Cost Benefit Analysis (CBA) constant Marginal Utility of Income (MUI) is usually assumed. This implies that a dollar received/earned is the same at low and high levels of income. In Social CBA, when comparing individuals, this implies that the analysis ignores distributions of income as a dollar accrued to a rich individual is just as "valuable" (in terms of the benefits) as a poor individual.
Are there examples you could come up with or papers that examine this assumption and look at the implications of using diminishing MUI's in place of constant MUI's? Would it just weight worse off individuals higher in the CBA analysis as they would be on the steeper part of the curve (assuming U(y) is the same for all individuals)?