The use of the compensation principle in cost-benefit analysis as a means of separating the efficiency and distributional effects of a project is theoretically suspect for a number of reasons. One difficulty is the lack of a necessary and sufficient criterion for determining that a compensation test is passed in an economy where consumers' and producers' prices are not the same. For this paper we establish such a criterion for projects that are small in the differential sense. The criterion requires the calculation of a set of "shadow prices" which is a weighted average of the consumers' and producers' price vectors.

The use of the compensation principle in cost-benefit analysis as a means of separating the efficiency and distributional effects of a project is theoretically suspect for a number of reasons. One difficulty is the lack of a necessary and sufficient criterion for determining that a compensation test is passed in an economy where consumers' and producers' prices are not the same. For this paper we establish such a criterion for projects that are small in the differential sense. The criterion requires the calculation of a set of "shadow prices" which is a weighted average of the consumers' and producers' price vectors.