| Abstract: | ABSTRACT. This paper presents a statistical procedure for empirically establishing linkages between regional economies. Using Granger's definition of causality and the final prediction error technique for lag-length specification suggested by Hsiao, a linked, multiregional employment model is developed. The forecast performance of the linked model is compared to models which do not include interregional linkages. It is shown that forecast error is reduced by taking into account employment interrelationships between regions. |
