Statistics seminar 2016: The Effects of Seasonal Adjustment Methods in Nonparametric Trend-Cycle Prediction

Seminario di Statistica.

  • Data: 04 maggio 2016 dalle 14:30 alle 16:00

  • Luogo: P.za Scaravili 2 - aula 3.

Contatto di riferimento:

Estela Bee Dagum is a former Full Professor of Time Series Analysis of the Faculty of Statistical Sciences of the University of Bologna where she was appointed by Chiara Fama. Previously she was the Director of the Time Series Research an Analysis Center of Statistics Canada, Ottawa, Canada. Professor Dagum is the main author of the X11ARIMA seasonal adjustment method , co-author of 20 books and more than 150 scientific articles published  in referees journals.

Abstract
The main reason for seasonal adjustment is the need of standardizing socioeconomic series because seasonality affects them with different timing and intensity. The information given by seasonally adjusted series has always played a crucial role in the analysis of current economic conditions and provides the basis for decision making and forecasting, being of major importance around cyclical turning points.

 Moving averages and ARIMA Model Based (AMB) methods are those mainly applied by statistical agencies to produce officially seasonally adjusted series. In recent years statistical agencies have shown interest in trend-cycle estimation, particularly  in  real time ,given the fact that socioeconomic series have  increased variability.

In a recent article published in the Annals of Applied Statistics, Dagum and Bianconcini introduced a new set of asymmetric filters  for real time trend cycle estimation that shows better predictability properties than the currently applied by statistical agencies.This seminar will discuss the basic assumptions of the two major seasonal adjustment methods applied by statistical agencies and their effects in nonparametric real time trend cycle estimation from the view points of revisions and time delay to detect a true turning point.