A methodology for aggregating industries of input-output models, with application to New Zealand interindustry data

Abstract

Aggregation, as it applies to input-output analysis, is the process of grouping industries of an interindustry model into sectors so as to produce an intersectoral model, which is more manageable in that its dimensions are smaller than those of the original model. A problem which arises as a result of aggregation is that, in general, forecasts of sector gross outputs, obtained from the intersectoral model, differ from those obtained by aggregating forecasts of industry gross outputs produced by the original interindustry model. This phenomenon is known as aggregation bias.

When there is a need to condense an input-output model into one of smaller dimensions, it is desirable that the resulting intersectoral model is subject to the smallest possible degree of aggregation bias. The researcher may also wish to place constraints upon the intersectoral model to ensure that certain industries are, or are not, aggregated into the same sector. Although there is a substantial body of theory which specifies the conditions under which the intersectoral model is not subject to aggregation bias, the extent to which various industry groupings satisfy these conditions is difficult to assess subjectively. Therefore, a formalised procedure, which groups industries into sectors so that aggregation bias is minimized, and which allows the imposition of constraints upon the intersectoral model, is a useful aid to the input-output analyst. Such a procedure is developed in this thesis and is applied to the most recent interindustry model of the New Zealand economy.