Specification and Testing of Analytic Techniques for Projecting State and Area Industry Employment

摘要

The principal objective of the research was to design and test the potential for improving the validity and accuracy of single-equation regression models for projecting local industry employment. The improved specification of such models would be based upon knowledge of specific characteristics and performance of the local economy in relationship to the national economy and knowledge of the product market and key interindustry linkages of a given industry. These analyses were conducted for 200 local economies in the U.S. and a representative sample of industries. A model specification framework was developed based upon generally accepted principles of regional economic theory. The selection of the appropriate set of predictor variables in a projection model for a given industry was based upon the results of the area and industry sector analysis. Projection models were calibrated for a large sample of industries in a variety of areas and states. The calibration results and a limited set of export projection tests indicate that the accuracy of single-equation regression models for projecting local industry employment can be improved significantly with more thoughtful pre-projection analyses and equation specifications.