Tracking the US Business Cycle With a Singular Spectrum Analysis

The monitoring of economic developments is an exercise of considerable importance forpolicy makers, namely, central banks and fiscal authorities as well as for other economic agents such as financial intermediaries, firms and households. However, the assessment of the business cycle is not an easy endeavor as the cyclical component is not an observable variable. In this paper we resort to singular spectrum analysis in order to disentangle the US GDP into several underlying components of interest. The business cycle indicator yielded through this method is shown to bear a resemblance with band-pass filtered output. As the end-of-sample behavior is typically a thorny issue in business cycle assessment, a real-time estimation exercise is here conducted to assess the reliability of the several filters. The obtained results suggest that the business cycle indicator proposed herein possesses a better revision performance than other filters commonly applied in the literature.