Investment Under Uncertainty: Theory and Tests with Industry Data

Under the assumption of constant returns to scale, there is a very simple, easily testable condition for optimal investment under uncertainty. Application of the test requires no parametric assumptions about technology and no assumptions about the competitiveness of the output market. The condition is that the expected marginal revenue product of labor equal the expected rental price of capital. The condition implies a certain invariance property for a modified version of Solow's productivity residual. Tests of the invariance property for U.S. industry data give very strong rejection in quite a few industries. The interpretation of rejection is either that the technology has increasing returns (possibly because of fixed costs) or that fins systematically over-invest.