Please use this identifier to cite or link to this item: http://hdl.handle.net/10419/104413
Authors: 
Blöchl, Andreas
Year of Publication: 
2014
Series/Report no.: 
Munich Discussion Paper 2014-3
Abstract: 
Penalized splines have become a popular tool to model the trend component in economic time series. The outcome of the spline predominantly depends on the choice of a penalization parameter that controls the smoothness of the trend. This paper derives the penalization of splines by frequency domain aspects and points out their link to rational square wave filters. As a novel contribution this paper focuses on the so called excess variability at the margins that describes the undesired increasing variability of the trend estimation to the ends of the series. It will be shown that the too high volatility at the margins can be reduced considerably by a time varying penalization, which yields more reliable estimations for the most recent periods.
Subjects: 
excess variability
penalized splines
spectral analysis
time varying penalization
trends
Persistent Identifier of the first edition: 
Document Type: 
Working Paper

Files in This Item:
File
Size





Items in EconStor are protected by copyright, with all rights reserved, unless otherwise indicated.