A Spectrum Analysis of the Long-Swing Hypothesis

Abstract

THE LONG-TERM ECONOMIC GROWTH of the American economy over the past century has frequently been characterized as expansion at a relatively constant rate. It is widely accepted that for some purposes this is a useful abstraction from the rather wide fluctuations observed in the annual growth rate. In more detailed studies of the growth of the American economy, a certain amount of interest has centered on the problem of isolating regular fluctuations about the long-term average value of the rate of expansion. Although some of the variation in the rate of growth is thought to be accounted for by the ordinary business cycle, several studies have suggested that the rate of growth accelerates and decelerates in a fairly regular pattern of some twenty years duration. These long swings are considered to be distinct from and independent of the shorter business cycle. This paper applies the technique of spectrum analysis to the problem of determining the statistical significance of long swings in the rate of growth of output and other related macro-economic variables. In the next section alternative methods of time series analysis are compared and the effect of low-pass filtering operations is considered. The empirical results of the study are presented in Section 3, and the main conclusions are summarized in the final section.