On choosing the optimal level of significance for the Durbin-Watson test and the Bayesian alternative

doi:10.1016/0304-4076(78)90029-5

Journal of Econometrics

Volume 8, Issue 2, October 1978, Pages 203-213

https://doi.org/10.1016/0304-4076(78)90029-5 Get rights and content

Abstract

This paper critically evaluates the usual ad hoc selection of the level of significance in the Durbin-Watson test and compares this procedure to the Bayesian alternative. The results of Monte Carlo experiments indicate that an α-level substantially larger than that normally used may be appropriate. The Bayesian estimator performed better than all preliminary test estimates in terms of MSE.

References (14)

P.A.V.B. Swamy et al.
Relative efficiencies of some simple Bayes estimators of coefficients in dynamic models—I
Journal of Econometrics
(1975)
T. Toyoda et al.
Estimation of variance after a preliminary test of homogeneity and optimal levels of significance for the pre-test
Journal of Econometrics
(1975)
J. Durbin et al.
Testing for serial correlation in least squares regression I
Biometrika
(1950)
J. Durbin et al.
Testing for serial correlation in least squares regression II
Biometrika
(1951)
J. Koerts et al.
On the power of the BLUS procedure
Journal of the American Statistical Association
(1968)
J.S. Mehta et al.
Efficient method of estimating the level of a stationary first-order autoregressive process
Communications in Statistics
(1974)
S.J. Prais et al.
Trend estimators and serial correlation
Unpublished Cowles Commission Discussion Paper Stat. no. 383
(1954)

There are more references available in the full text version of this article.

Cited by (44)

Asymptotic behaviour of regression pre-test estimators with minimal Bayes risk
2007, Journal of Econometrics
Why are estimates of agricultural supply response so variable?
1997, Journal of Econometrics
Estimates of the response of agricultural supply to movements in expected price display curiously large variation across crops, regions, and time periods. We argue that this anomaly may be traced, at least in part, to the statistical properties of the commonly-used econometric estimator, which has infinite moments of all orders and may have a bimodal distribution. We propose an alternative minimum-expected-loss estimator, establish its improved sample properties, and argue for its usefulness in the empirical analysis of agricultural supply response.
Higher moment estimators for linear regression models with errors in the variables
1997, Journal of Econometrics
This paper proposes consistent instrumental variable estimators for linear regression models with errors in the variables that require no extraneous information. These estimators are based on sample moments of order higher than two. While similar estimators proposed previously in the literature seem to be quite erratic, our experimental f findings suggest that our estimators perform better than ordinary least squares estimators in terms of root mean squared errors and also in terms of size of type I errors of standard tests in many typical situations of economic analyses. Tests for the presence of errors in the variables are also described.
Comments on testing economic theories and the use of model selection criteria
1995, Journal of Econometrics
This paper outlines difficulties with testing economic theories, particularly that the theories may be vague, may relate to a decision interval different from the observation period, and may need a metric to convert a complicated testing situation to an easier one. We argue that it is better to use model selection procedures rather than formal hypothesis testing when deciding on model specification. This is because testing favors the null hypothesis, typically uses an arbitrary choice of significance level, and researchers using the same data can end up with different final models.
Bayes regression with autoregressive errors. A Gibbs sampling approach
1993, Journal of Econometrics
This paper develops a practical framework for the Bayesian analysis of Gaussian and Student-t regression models with autocorrelated errors. As is customary in classical estimation procedures, the posteriors are conditioned on the initial observations. Recourse is taken to the method of Gibbs sampling, an iterative Markovian sampling method, and it is shown that the proposed approach can readily deal with high-order autoregressive processes without requiring an importance sampling function or other tuning constants. Several examples, including one with AR(4) errors, are used to illustrate the ideas.
Fighting the teflon factor. Comparing classical and Bayesian estimators for autocorrelated errors
1991, Journal of Econometrics
This paper claims, on the basis of a Monte Carlo study, that current estimators employed in the context of first-order autocorrelated errors should be abandoned in favor of a Bayesian alternative. Frequentists are confronted on their own playing field: The Bayesian estimator employs an ignorance prior and the criterion adopted is the mean square error of the competing estimators. With the predictable exception of the OLS estimator when the autocorrelation parameter is close to zero, the Bayesian estimator dominates the competition by a substantive margin.

View all citing articles on Scopus

View full text

On choosing the optimal level of significance for the Durbin-Watson test and the Bayesian alternative

Abstract

Journal of Econometrics

Journal of Econometrics

Testing for serial correlation in least squares regression I

Biometrika

Testing for serial correlation in least squares regression II

Biometrika

On the power of the BLUS procedure

Journal of the American Statistical Association

Efficient method of estimating the level of a stationary first-order autoregressive process

Communications in Statistics

Trend estimators and serial correlation

Unpublished Cowles Commission Discussion Paper Stat. no. 383