Long-Run Production Modeling with Pseudo Data: Electric Power Generation

This paper presents a new approach to long-run production modeling which combines the simplicity of the statistical cost function with the technical detail of process analysis. Pseudo data, which are generated by an electric power process model, depict the cost-minimizing input configurations for alternative relative input prices. The pseudo data are then utilized to estimate a translog cost function, which provides price and substitution elasticities as well as a convenient form for micromodeling. Pseudo data offer numerous advantages compared to conventional time series particularly in that they avoid multicollinearity, a limited sample range, and inadequate technical and environmental detail.