Estimation of Process Change for Industrial Pollution Abatement

Abstract

When forecasting the impacts of pollution control regulations upon industries which produce pollutants jointly with ordinary outputs, potential input and production process adjustments must be identified and assessed. The problem addressed here is that the empirical approximation of Joint production processes by a single-equation Least Squares (LS) regression misrepresents restrictions on input substitution possibilities. In such cases, some inputs are allocated to both production processes, which provides the jointness. But other inputs contribute to only one process and do not contribute to adjustment opportunities in the other process. Since single-equation LS regression assumes that all inputs can be substituted throughout the production process, its application results in a misspecified empirical model. This paper presents a theoretical framework that motivates the use of Seemingly Unrelated Regression (SUR) estimators to directly approximate such processes. The advantage to using the SUR estimators is that they exploit the related nature of the input decisions so that a more accurate model of the processes can be estimated. The framework is demonstrated with an application to the case of SO₂ abatement by coal-fired electricity generating units. The results indicate that these units could efficiently burn more low-BTU, low-sulfur coal to maintain the same electricity output with a lower SO₂ emission rate. More generally, this application demonstrates the usefulness of the framework developed here.