|
Activity Number:
|
130
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 3, 2009 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Business and Economic Statistics Section
|
| Abstract - #303974 |
|
Title:
|
Efficiently Forecasting Thousands of Employment Series for Sub-State Areas
|
|
Author(s):
|
David E. Byun*+ and Thomas Evans
|
|
Companies:
|
Bureau of Labor Statistics and Bureau of Labor Statistics
|
|
Address:
|
2 Massachusetts Avenue NE, Washington, DC, 20212,
|
|
Keywords:
|
ARIMA ; Forecast ; Small Area
|
|
Abstract:
|
This poster presentation describes an automated system that efficiently forecasts thousands of time series. Our system is applied to payroll employment data that are important inputs to the development of unemployment rates for small areas. The data source is the Bureau of Labor Statistics' Quarterly Census of Employment and Wages, which produces counts of payroll employment at the county level, and is only available after a lag of six months. Our approach has increased efficiency while still yielding good forecasts for the more than 2,400 Labor Market Areas in our system. Initial ARIMA models are chosen by an automatic modeling and outlier detection procedure in X-12 ARIMA that provides satisfactory results for most series. For series with inadequate models, various diagnostics including spectrum plots are used to attempt to improve on the automatic modeling results.
|
