|
Activity Number:
|
444
|
|
Type:
|
Invited
|
|
Date/Time:
|
Wednesday, August 5, 2009 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Business and Economic Statistics Section
|
| Abstract - #302760 |
|
Title:
|
Recent Developments in Trend-Cycle Prediction for Real Time Analysis
|
|
Author(s):
|
Estela Bee Dagum*+ and Silvia Bianconcini
|
|
Companies:
|
University of Bologna and University of Bologna
|
|
Address:
|
via delle Belle Arti 41, Bologna, 40126, Italy
|
|
Keywords:
|
polynomial kernel regression ; smoothing cubic splines ; spectral properties ; real time analysis ; revisions
|
|
Abstract:
|
Financial, economic and environmental globalization have introduced a large amount of volatility in seasonally adjusted data which can hardly be used to assess the short-term trend direction. The purpose of this study is to present a unified probabilistic view of nonparametric trend-cycle predictors via Reproducing Kernel Hilbert Space methodology. We discuss symmetric and non-symmetric filters for real time analysis. Applied to a large sample of series, it is shown that the Kernel predictors give better results than the classical ones in terms of signal passing, noise suppression, and revisions when new observations are added to the series.
|
