[Test de la causalité instantanée linéaire de Granger en présence de dynamiques non linéaires]
Dans cette Note on considère le test de la causalité instantanée linéaire au sens de Granger entre deux variables dans le cas où les innovations sont dépendantes mais non corrélées (cʼest-à-dire des innovations linéaires). Les hypothèses considérées sont faibles et peuvent prendre en compte des non linéarités comme par exemple celles induites par les processus GARCH. Nous établissons que la statistique de Wald standard pour tester la causalité instantanée linéaire nʼest pas valide dans notre cadre. En conséquence des tests de Wald valides sont proposés.
This Note is devoted to the test of instantaneous linear Granger causality when the errors are dependent but uncorrelated. The assumptions are weak and include a large set of dynamics as for instance the GARCH processes. We show that the standard Wald test for testing instantaneous linear Granger causality is not valid in our framework. As a consequence Wald tests which are valid in our framework are proposed.
Accepté le :
Publié le :
Hamdi Raïssi 1
@article{CRMATH_2011__349_21-22_1203_0, author = {Hamdi Ra{\"\i}ssi}, title = {Testing instantaneous linear {Granger} causality in presence of nonlinear dynamics}, journal = {Comptes Rendus. Math\'ematique}, pages = {1203--1206}, publisher = {Elsevier}, volume = {349}, number = {21-22}, year = {2011}, doi = {10.1016/j.crma.2011.10.008}, language = {en}, }
Hamdi Raïssi. Testing instantaneous linear Granger causality in presence of nonlinear dynamics. Comptes Rendus. Mathématique, Volume 349 (2011) no. 21-22, pp. 1203-1206. doi : 10.1016/j.crma.2011.10.008. https://comptes-rendus.academie-sciences.fr/mathematique/articles/10.1016/j.crma.2011.10.008/
[1] Maximum likelihood estimation for all-pass time series models, Journal of Multivariate Analysis, Volume 97 (2006), pp. 1638-1659
[2] An improved heteroskedasticity and autocorrelation consistent covariance matrix estimator, Econometrica, Volume 60 (1992), pp. 953-966
[3] Stochastic comparison of tests, Annals of Mathematical Statistics, Volume 31 (1960), pp. 276-295
[4] Mixing and moment properties of various GARCH and stochastic volatility models, Econometric Theory, Volume 18 (2002), pp. 17-39
[5] Multivariate portmanteau test for autoregressive models with uncorrelated but nonindependent errors, Journal of Time Series Analysis, Volume 28 (2007), pp. 454-470
[6] Diagnostic checking in ARMA models with uncorrelated errors, Journal of American Statistical Association, Volume 100 (2005), pp. 532-544
[7] Computing the distribution of quadratic forms in normal variables, Biometrika, Volume 48 (1961), pp. 419-426
[8] New Introduction to Multiple Time Series Analysis, Springer, Berlin, 2005 (pp. 41–43)
Cité par Sources :
Commentaires - Politique