The seasonal adjustment of economic data by spectral methods

Abstract

The object of this thesis is to present a formal theory for the seasonal adjustment of a single series in terms of spectral analysis, that is, the analysis of frequency components, together with a practical illustration. Instead of describing the well-know classical procedures we refer the reader to two quite sophisticated examples, the work of the Bank Deutscher Lander (1957) and Shiskin and Eisenpress (1957). It will appear that the spectral method is more powerful for description and analysis than are the classical methods, since it decomposes variables into their fundamental component parts; the estimation of variances and tests of significance are thus greatly facilitated. (First paragraph of 1.2)