|
Activity Number:
|
510
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Wednesday, August 5, 2009 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Business and Economic Statistics Section
|
| Abstract - #304236 |
|
Title:
|
An Empirical Evaluation of Signal Extraction Goodness-of-Fit Diagnostic Tests
|
|
Author(s):
|
Christopher Blakely*+ and Tucker S. McElroy
|
|
Companies:
|
U.S. Census Bureau and U.S. Census Bureau
|
|
Address:
|
Silver Hill Road, Washington, DC, ,
|
|
Keywords:
|
Time Series ; Signal Extraction ; ARIMA Models
|
|
Abstract:
|
We present a suite of computational Monte Carlo studies of a band-limited frequency-domain goodness-of-fit (gof) diagnostic test for signal extraction in stationary time series. One of the novel features of this diagnostic test is that it takes into account the effects of model parameter uncertainty which then aims at providing a consistent estimate of the variance of the gof diagnostic statistics. We show computationally that this effectively allows us to obtain adequate size and power properties of the gof diagnostic statistics for a relatively small number of observations in the time series. We provide complete documentation of the numerical properties of the diagnostics through Monte Carlo studies of finite sample size and power for different combinations of both signal and noise components using canonical seasonal, trend, and irregular components.
|
