The Performance of Alternative Integration Strategies for Combining Regional Econometric and Input-Output Models

Abstract

As the number of integrated econometric and input-output models appearing in the literature grows, the need for a closer examination of the different approaches towards integration is warranted. This paper suggests a taxonomy of strategies for combining econometric and input-output models at the regional scale. The properties of the different integration strategies are then examined from a theoretical perspective, focusing on the treatment of model closure, dynamic representation. interregional linkages, sectoral aggregation, and estimation issues. Initial empirical insight as to the simulation and impact analysis capabilities of the alternative integration strategies is provided through a series of experiments employing integrated multiregional econometric and input-output models for five regions in Southern California. The results suggest that integrated models as a class can provide for improved forecasting accuracy as well as more comprehensive impact analysis capabilities over traditional non-integrated models. Additional insights are provided regarding the sensitivity of model performance to the choice of integration strategy.