Title: Factor models by frequency by multivariate circulant singular spectrum Analysis - MCSSA
Authors: Pilar Poncela - European Comission Joint Research Centre (Italy)
Juan Bogalo - University of Alcala (Spain)
Eva Senra - Universidad de Alcala (Spain) [presenting]
Abstract: Circulant Singular Spectrum Analysis (CSSA) allows the extraction of the unobserved components associated with any ex-ante desired frequency in a time series in an automatic way. We first generalize the technique to a multiple setup and automatize it in the same way by the use of circulant matrices applied to the new multivariate trajectory matrix. Second, we extend our proposed methodology to perform factor analysis by frequency. In this way we can jointly extract common factors associated to the trend, to the cycle or to the seasonal component. Finally, we show the application of MCSSA to find common cycles in economic data by sectors or countries.