Resumo:
This paper aims to show that the use of multivariate analysis facilitates comparative static studies on the economy from a sectorial approach. An algorithm of cluster analysis, called affinity propagation, was used to find some temporal patterns in the multipliers in the US economy between 1997 and 2017. The study enabled us to reach three main conclusions. First, to confirm that the established concept that structural changes in the US economy did not occur suddenly. Second, the five-year interval time (as upper limit) may be inappropriate for studies based on input-output analysis for the US economy in the period. Third, the formation of the clusters (periods with structural similarities) of both backward and forward multipliers was sensitive to the 2007 Financial Crisis.
