Another View of the Long Swing: Comments on Adelman's Study of Long Cycles

摘要

Prof. Irma Adelman's recent study, 'Long Cycles Fact or Artifact.' ('American Economic Review,' June 1965, p. 444-463), represents an attempt to apply spectral analysis to time series data in order to ascertain the relative contribution of a long-swing component to the historical time path of economic activity in the United States. This paper has been prepared in an attempt to point out some of the subleties involved in the application of spectral analysis and to suggest that Adelman's rejection of the long-swing hypothesis is not well-founded. Although the paper is concerned primarily with the long-swing hypothesis, the comments on hypothesis testing and filtering techniques are applicable to a wider class of problems and illustrate the flexibility of the estimation technique. The basic conclusion which emerges is that the estimates presented by Adelman contain virtually no information about the long swing. In an attempt to rectify this deficiency, two alternative estimation procedures are introduced. The alternative estimates which are obtained suggest that the case for the long-swing hypothesis is mixed; some of the evidence examined being favorable and some unfavorable. Without further analysis it is impossible to formulate a definite conclusion about the existence of long cycles in the historical pattern of development of the United States economy. (Author)