Application of spectral methods in economic data analysis

Abstract

Because a stationary time series may be represented in spectral (or frequency) terms it has become apparent that certain areas of economic investigation can be effectively performed in this domain. This approach may arise because a priori information is most easily expressed in frequency terms or it may be that greater insight results from using spectral methods in conjunction with the more usual procedures developed for the time domain. A large area of empirical economics exists where data analysis in the frequency domain is not contemplated because of insufficient data. Nevertheless, where theoretical explanations are sought spectral methods may still be very useful.