Revisiones en el ciclo económico con Circulant SSA
- Juan Vicente Bógalo Román
- Pilar Poncela
- Eva Senra
- Cristina Tejedor Martínez (coord.)
- Antonio Guerrero Ortega (coord.)
- Germán Ros Magán (coord.)
- Francisco Pascual Vives (coord.)
- Paloma Ruíz Benito (coord.)
- Vanessa Tabernero Magro (coord.)
Publisher: Editorial Universidad de Alcalá ; Universidad de Alcalá
ISBN: 978-84-16599-47-9
Year of publication: 2017
Pages: 187-198
Congress: Jornadas de Jóvenes Investigadores de la Universidad de Alcalá (6. 2017. null)
Type: Conference paper
Abstract
Singular Spectrum Analysis, SSA, is a powerful non-parametric technique of signal extraction in time series. The Circulant version stands out for its good separability properties of the estimated components, both with stationary and non-stationary time series. In this paper we analyze the effect of the size of the window on the revisions of the estimated business cycle using as estimator the first difference of the trend-cycle. The analysis is performed with real series such as the IPI of France and USA where the revisions produced by Circulant SSA are compared with those obtained by a parametric procedure based on ARIMA models, such as Tramo-Seats. It is noted that Circulant SSA produces smaller revisions.