Solving the Problem of Socially-Improving Multivariate Tax Reform with s-order Stochastic Dominance: An Application to Egyptian Consumption

Abstract

This paper integrates two pans of the operations research literature, optimal tax design and stochastic dominance analysis, to characterize indirect tax reforms that improve welfare or reduce poverty. We develop a general multivariate framework based on s-order stochastic dominance that extends existing approaches beyond the traditional two-good setting and beyond second-order dominance. The proposed method identifies budget-neutral tax reforms that are robust for broad classes of social welfare functions. Tax reforms are derived from a convex optimization problem and assessed with a wild bootstrap test adapted to dominance conditions. An application to Egyptian household data illustrates the empirical relevance of the approach and underscores the importance of higher-order dominance criteria for designing socially improving tax policies.