|
Activity Number:
|
405
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Wednesday, August 9, 2006 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Business and Economics Statistics Section
|
| Abstract - #306565 |
|
Title:
|
An ARIMA Model--Based Approach To Estimate Evolving Trading Day Effect
|
|
Author(s):
|
Xichuan Zhang*+ and Anna Poskitt
|
|
Companies:
|
Australian Bureau of Statistics and Australian Bureau of Statistics
|
|
Address:
|
ABS House, Belconnen ACT, 2617, Australia
|
|
Keywords:
|
trading day effect ; seasonal adjustment ; ARIMA ; random coefficient
|
|
Abstract:
|
An important aspect of the seasonal adjustment of monthly economic flow time series is the estimation and correction for nonregular calendar variation, termed trading day effects. A static trading day effect often is assumed and estimated based on a regression-ARIMA framework in seasonal adjustment software such as X-12-ARIMA and TRAMO. However, this assumption is not always realistic. To improve seasonal adjustment quality within the regression-ARIMA framework, this paper presents a method utilizing a rolling window time-series span in conjunction with various smoothing techniques to estimate an evolving trading day effect. Quality assessment is made for the proposed method in comparison with other methods suggested in literature using simulated time series. The proposed method also is evaluated using Australian Bureau of Statistics monthly time series.
|
