|
Activity Number:
|
474
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 1, 2007 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Business and Economics Statistics Section
|
| Abstract - #308310 |
|
Title:
|
Inferences in Regression with Correlated Residuals
|
|
Author(s):
|
Yue Fang*+
|
|
Companies:
|
University of Oregon
|
|
Address:
|
CEIBS, College of Business, Eugene, OR, 97403,
|
|
Keywords:
|
Autocorrelation ; Hypothesis test ; Generalized least squares ; Regression
|
|
Abstract:
|
In this article we show that autoregressive corrections based on generalized least squares (GLS) estimation yields estimates for serially correlated regression parameters that are asymptotically efficient, and that have the same limiting distribution as those of Aitken estimators (GLS). Based on exhaustive Monte Carlo studies we also demonstrate that for finite samples the relative efficiency this estimator is comparable to that of estimated GLS when the autoregressive order is appropriately chosen. Moreover, this type of correction can yield as good, and more often, better forecasts than those generated from ordinary least squares (OLS) or from GLS using the correct form of the residual autocorrelation structure. This suggests that there is not much to be gained in trying to identify the correct form and order of the serial correlation.
|
- 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 2007 program |
