Variable cost functions and the rate of return to quasi-fixed factors: an application to r and d in the Bell system

We formulate a variable cost function model in which certain inputs are treated as quasi-fixed, and develop a simple statistical test of whether optimization occurs for the quasi-fixed inputs. It is shown how to retrieve characteristics of the long-run cost function from the variable cost parameters, with specific reference to the cost elasticity and the elasticities of substitution. We also present a model of the I returns to R & D in the context of a regulated firm and show how to I estimate the net rate of return to R & D from the variable cost function. A translog version of the model is estimated for the Bell System for the period 1947-1976. The empirical results suggest substantial long-run economies of scale at the aggregate level. The formal envelope test indicates that the Bell System's use of capital and R & D was cost- minimizing during the post-war period, but the conclusion is seriously qualified by evidence that the power of the test in this application is low. Finally, we estimate the net rate of return to R & D in the Bell System in the range of 25-40 percent, which is somewhat higher than available estimates for manufacturing industries.