Decomposition of the European GDP based on Singular Spectrum Analysis

In this paper, the Singular Spectrum Analysis (SSA), a relatively new tool originated in natural sciences, for orthogonal decomposition of time series, is presented and applied in the European real, seasonally unadjusted quarterly GDP for the period 1995 - 2010. SSA is suitable for short and noisy time series, properties that characterize many macroeconomic time series. In this paper, I decompose the GDP in trend, cycle, seasonals and noise components. There are significant similarities but also some differences between the SSA-based filter and the other well-known macroeconomic filters. These differences are shown here by means of correlation matrices and spectral measures. Although SSA is a method that only very recently has been introduced in macroeconomics, its use in the natural sciences for more than three decades, makes it a serious candidate for tackling macroeconomic issues such as filtering, denoising and smoothing.