Statistics and Algorithms / GARCH Models / GARCH Models (Univariate) /

UV GARCH GARCH-X

If you add other explanatory variables to your variance model besides those formed from lagged residuals and lagged variances, you get an ARCH-X or GARCH-X model. To do this, use the option XREGRESSORS and add a supplementary card listing the extra explanatory variables in the variance equation. (Don’t include CONSTANT, which will always be in the variance model).

If you need both regressors for the mean model and for the variance model, put the card for the mean model regressors first. Suppose you have a dummy variable for Mondays (note that there isn't one in this data set), and you think that the variance is systematically different on Mondays. This adds the “Monday” effect to the variance model:

garch(p=1,q=1,regressors,xreg) / dlogdm

# constant dlogdm{1}

# monday