Sufficient Statistics Revisited

This article reviews and generalizes the sufficient statistics approach to policy evaluation. The idea of is that welfare effect changes can be expressed in terms estimable reduced-form elasticities, allowing for evaluation without estimating structural primitives fully specified models. relies on three assumptions: are small, government only source market imperfection, a set high-level restrictions environment preferences used reduce number elasticities estimated. We generalize all dimensions. It possible develop transparent formulas under very general conditions, but estimation requirements increase greatly. Starting from such elucidates feasible empirical implementations fact approaches.