Spatial Aggregation in Multiregional Input-Output Models

Abstract

This study continues the investigations into effects of spatial aggregation in connected, several-region input-output models that were initially reported upon in Miller and Blair (1980; 1981) for an interregional input-output model. In the present paper our interest is in the error that is introduced into a multiregional input-output model through aggregation of regions. As in the earlier paper, we investigate the question by means of both hypothetical and real data. In the present study, these are (1) randomly generated multiregional data incorporating varying degrees of regional linkage and (2) actual 1963 US multiregional data. To allow some comparison with the interregional experiments, the data for the random experiments in the present paper are derived from our original randomly generated interregional data. The same levels of spatial interconnectedness as in the interregional experiments are also used here, and the same measures of aggregation error are employed. As in the interregional case, we conclude that spatial aggregation in multiregional input-output models does not necessarily lead to unacceptable error.