How should we measure economic efficiency? The canonical measure is an unweighted sum of willingnesses to pay. In contrast, this paper provides efficient welfare weights that implement the Kaldor-Hicks tests for efficiency but account for the distortionary cost of taxation. The shape of the income distribution yields bounds on these weights that suggest it is efficient to weight surplus to the poor more than to the rich. Point estimates suggest surplus to the poor should be weighted 1.5-2x more than surplus to the rich. I illustrate how to use these weights to evaluate the efficiency of government policy changes.
(An earlier version circulated under the title "The Inequality Deflator: Interpersonal Comparisons without a Social Welfare Function", available here)