ENERGY SUBSTITUTION IN U. S. MANUFACTURING AND AGRICULTURE

This thesis investigates the relationship of energy to other nonenergy inputs, specifically, captial, labor and intermediate materials in twenty manufacturing sectors from 1947 to 1976. To do this, two models are used, a nonlinear static model and a dynamic linear model. The functional form of the nonlinear model is a generalized Box-Cox cost function. The form of the dynamic linear model is a vector autoregression with an exchangeability prior.Because of cost constraints in the maximum likelihood estimation of the GBC cost function, only five sectors are analyzed these are: paper products, chemical and allied products, petroleum and coal products, primary metals and agriculture.The important results from this estimation show that capital and energy are substitutes and labor and energy are complements in paper products, primary metals and agriculture. The opposite relationships exist in chemical products and petroleum products. All sectors show that technical change is labor saving, energy and capital using. Primary metals and agriculture show technical change is materials using but paper products, chemical products and petroleum and coal show that technical change is materials saving. Increases in energy and capital price also have a negative effect on total factor productivity in all five sectors.The linear vector autoregressive model incorporates an exchangeability prior. This prior assumes coefficients from each sector have the same distribution. This prior yields an estimation procedure that utilizes all of the data, yet allows separate sectoral estimates to be obtained, unlike more restrictive pooled data procedures. These estimates are obtained with an interative procedure. The variables used in this model are output, capital, price of capital, labor, price of labor, energy, price of energy, materials and price of materials.The generalized Box-Cox cost function (GBC) allows estimation of elasticities of substitution, price elasticities of input demand, economies of scale and bias to technical change without a priori restriction. The elasticity of total factor productivity is also obtainable from the estimated parameters.The vector autoregressive model has associated with it an impulse response function. This impulse response function gives the response of all variables to a change in any one variable. This framework is particularly useful in determining the response of the system to an increase in energy prices. This response is dynamic and quantifies in a step by step process how output, inputs and prices adjust to energy price increases.The important conclusions from this model show that in most sectors capital increases in response to energy price increases, but this captial increase is not sustained, indicating that capital purchases are geared more towards one time conservation measures rather than extensive changes in the production process. Only rubber and miscellaneous plastics products show no increase in the capital stock. These results show that in the long run, growth in output may be somewhat hindered by energy price increases.Policy suggestions made include a BTU tax, as this taxes energy types and energy users fairly, it also stimulates conservation. It is also recommended that the pool of funds this tax creates be used for research into new production processes as well as alternative energy sources so that negative effects of energy price increases on output would be minimized.