This paper explores the extension of integrated econometric/input-output models from a single to a multiregional context. A number of methodological issues associated with existing treatments of extraregional relations in both traditional multiregional econometric models and the more recent single region integrated models are discussed. A series of new approaches are introduced that rely on the integration of econometric and input-output methods to specify extraregional linkages. A comparison of these alternative modeling approaches is undertaken using employment series for 5 regions in Southern California. The forecasting accuracies are evaluated in a series of out-of-sample forecasting experiments. The findings suggest that the forecasting performance of the multiregional extensions of the integrated approach is competitive with the performance of single region implementations. The performance of the alternative multiregional approaches is found to be relatively more sensitive to regionalization of the IO component and the level of industrial disaggregation than to the incorporation of distance in the specification of the multiregional linkages.
|Original language||English (US)|
|Number of pages||22|
|Journal||Growth and Change|
|State||Published - Jan 1 1997|
ASJC Scopus subject areas
- Global and Planetary Change