| Sumario: | The article focuses on effects of parametric uncertainty and technological change on input-output models in the U.S. Parameters of input-output (I-O) models are not known exactly; they are subject to two major types of uncertainty. First, statistical errors in compiling the massive data base can introduce uncertainty into all analyses, even those for the base year; and second, uncertainty arises from the fact that I-O coefficients do not remain constant over time, violating an assumption made in most forecasting applications. Since their introduction over thirty years ago, most I-O models have been based on data necessarily aggregated to less than 150 sectors because of computational limitations. Recently, however, it has become possible to manipulate models having as many as 300 to 500 sectors. The problem of parametric forecasting has been the subject of much analysis since the Cambridge Growth Project (1963) introduced the disproportional RAS method of adjusting I-O coefficients to be consistent with updated control totals. In this paper, authors demonstrate a method for restructuring I-O models so parameters more accurately reflect real technological options, a framework in which parametric forecasting may be done with greater certainty.
|
|---|