|
Activity Number:
|
108
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Monday, August 9, 2004 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Business and Economics Statistics Section
|
| Abstract - #302204 |
|
Title:
|
Seasonal Adjustment of Weekly Time Series with Application to Unemployment Insurance Claims and Steel Production
|
|
Author(s):
|
William P. Cleveland*+
|
|
Companies:
|
Federal Reserve Board
|
|
Address:
|
, Washington, DC, ,
|
|
Keywords:
|
|
|
Abstract:
|
New seasonal patterns in unemployment insurance claims have prompted a change in the method of seasonal adjustment. Seasonal adjustment of weekly series at the Federal Reserve and of initial unemployment claims at the Bureau of Labor Statistics was accomplished until recently by fixed coefficient regression. The seasonal variables were trigonometic expansions with periods 365 (366 in leap years) and harmonics to model the effect of a week ending on a particular day of the year. These variables capture the changing position of weekdays from one year to the next. While we continue to estimate holiday effects as stable over extended periods, estimation of the rest of the seasonal pattern permits an evolution of the pattern from year to year beyond what method suggested takes advantage of the periodicity of the seasonal desgin matrix to reduce the amount of computation required. Weighting schemes to achieve parameter estimates are determined by signal extraction formulas set to achieve a rate of adaptation dictated by appropriate objective functions.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2004 program |
