2010
6/2010
Long cycles in growth: explorations using new frequency domain techniques with US data
Author(s): Patrick M Crowley
2010. 49 pages.
Publisher: Bank of Finland
ISBN 978-952-462-586-9 (Printed publication)
ISBN 978-952-462-587-6 (Web publication)
Search words: business cycles, growth cycles, frequency domain, spectral analysis, long cycles, Granger, wavelet analysis, Hilbert-Huang Transform (HHT), empirical mode decomposition (EMD), non-stationarity, C13, C14, O47, Patrick M. Crowley
In his celebrated 1966 Econometrica article, Granger first hypothesized that there is a ‘typical’ spectral shape for an economic variable. This ‘typical’ shape implies decreasing levels of energy as frequency increases, which in turn implies an extremely long cycle in economic fluctuations and particulary in growth. Spectral analysis is however based on certain assumptions particulary in that render these basic frequency domain techniques inappropriate for analysing non-stationary economic data. In this paper three recent frequency domain methods for extracting cycles from non-stationary data are used with US real GNP data to analyse fluctuations in economic growth. The findings, among others, are that these more recent frequency domain techniques do not provide evidence to support the ‘typical’ spectral shape and nor an extremely long growth cycle á la Granger.