On the theory of DRC criteria

Srinivasan and Bhagwati (1978) show total and direct DRC are identical in a two-good, two-factor model with an additional, small project. This result is shown here not to extend to the case of more goods than factors, the usual empirical context of DRCs. For first-best solutions the direct DRC is appropriate, the total DRC is biassed against production and the so-called ideal variant biassed toward production. The total DRC is often advocated when areas of control are limited. If production or trade constraints compel obtaining materials from domestic producers, a particular, total DRC decomposition does provide valid guidelines, but its measure requires prior knowledge of the solution. The direct DRC, measured in second-best shadow prices, does convey limited information, even when areas of control are restricted, but in general a total formulation diminishes this information content.