zbMATH — the first resource for mathematics

Estimation and hypothesis testing of cointegration vectors in Gaussian vector autoregressive models. (English) Zbl 0755.62087
Three basic issues in Gaussian vector autoregressive (VAR) models with a mean drift and seasonal dummy variables are discussed: first, the problem of finding the number of cointegrating relations in nonstationary data; second the problem of statistically estimating the cointegrating vectors, and finally that of statistically testing submodels specifying interesting economic hypotheses about their structure. The VAR model with intercept and seasonal dummies is analyzed by maximum likelihood methods. The likelihood method is compared with several other alternative methods such as the nonparametric spectral regression method due to P. C. P. Phillips [see: Nonparametric and semiparametric methods in econometrics and statistics, Proc. 5th Int. Symp., Econ. Theory Econ., Durham/NC (USA) 1988, 413-435 (1991)] and a three stage estimator for the error correction model due to R. F. Engle and B. S. Yoo [see J. Econ. 35, 143-159 (1987; Zbl 0649.62108)]. However no robustness tests are performend here.

62P20 Applications of statistics to economics
62M10 Time series, auto-correlation, regression, etc. in statistics (GARCH)
Zbl 0649.62108
Full Text: DOI Link