Some tests for parameter constancy in cointegrated VAR-models
Research output: Contribution to journal › Journal article › Research › peer-review
Standard
Some tests for parameter constancy in cointegrated VAR-models. / Hansen, Henrik; Johansen, Søren.
In: Econometrics Journal, Vol. 2, No. 2, 1999, p. 306-333.Research output: Contribution to journal › Journal article › Research › peer-review
Harvard
APA
Vancouver
Author
Bibtex
}
RIS
TY - JOUR
T1 - Some tests for parameter constancy in cointegrated VAR-models
AU - Hansen, Henrik
AU - Johansen, Søren
PY - 1999
Y1 - 1999
N2 - Some methods for the evaluation of parameter constancy in vector autoregressive (VAR) models are discussed. Two different ways of re-estimating the VAR model are proposed; one in which all parameters are estimated recursively based upon the likelihood function for the first observations, and another in which the cointegrating relations are estimated recursively from a likelihood function, where the short-run parameters have been concentrated out. We suggest graphical procedures based on recursively estimated eigenvalues to evaluate the constancy of the long-run parameters in the model. Specifically we look at the time path of the eigenvalues using a new result on the asymptotic distribution of the estimated eigenvalues. Furthermore we show that the fluctuation test by Ploberger et al. (1989) and the Lagrange multiplier (LM) type test for constancy of parameters by Nyblom (1989) can be applied to test the constancy of the long-run parameters in the cointegrated VAR-model. All results are illustrated using a model for the term structure of interest rates on US Treasury securities.
AB - Some methods for the evaluation of parameter constancy in vector autoregressive (VAR) models are discussed. Two different ways of re-estimating the VAR model are proposed; one in which all parameters are estimated recursively based upon the likelihood function for the first observations, and another in which the cointegrating relations are estimated recursively from a likelihood function, where the short-run parameters have been concentrated out. We suggest graphical procedures based on recursively estimated eigenvalues to evaluate the constancy of the long-run parameters in the model. Specifically we look at the time path of the eigenvalues using a new result on the asymptotic distribution of the estimated eigenvalues. Furthermore we show that the fluctuation test by Ploberger et al. (1989) and the Lagrange multiplier (LM) type test for constancy of parameters by Nyblom (1989) can be applied to test the constancy of the long-run parameters in the cointegrated VAR-model. All results are illustrated using a model for the term structure of interest rates on US Treasury securities.
KW - Faculty of Social Sciences
KW - cointegration
KW - eigenvalues
KW - fluctuation test
U2 - 10.1111/1368-423X.00035
DO - 10.1111/1368-423X.00035
M3 - Journal article
VL - 2
SP - 306
EP - 333
JO - Econometrics Journal
JF - Econometrics Journal
SN - 1368-4221
IS - 2
ER -
ID: 81850