Structural Decomposition Analysis

When there are two or more sets of Input–Output data for an economy, analysts are often interested in trying to disaggregate the total amount of change in some aspect of that economy into contributions made by its various components. For example, the total change in gross outputs between two periods could be broken down into that part associated with changes in technology (as reflected, initially, in the changes in the Leontief inverse for the economy over the period) and that part related to changes in final demand over the period.

At the next level, the total change in the Leontief inverse matrix could be disaggregated into a part that is associated with changes in technology within each sector (as reflected in changes in the direct input coefficients matrix) and that part associated with changes in product mix within each sector. Similarly, the change in final demand could be further disaggregated into a part that reflects changes in the overall level of final demand and a part that captures changes in the composition of final demand. And there are numerous additional options – for example, there is no need to use only two contributing factors; changes in employment, value added, energy use, etc. may be of more economic interest than changes in gross outputs; and so on. For a general overview of this literature, see Rose and Casler (1996) or Dietzenbacher and Los (1997, 1998). Two early empirical examples of this kind of work can be found in Feldman, McClain and Palmer (1987) for the USA and Skolka (1989) for Austria.