The RATS Software Forum

Dear all,

I employed the multivariate VAR(1)-DCC-GARCH(1,1) model in empirical analysis. I wonder if the standard errors are avaialable in the GIRF
(Generalized Impulse Response Functions). If yes, could I have the code for version pro 8.1. If no, why? How about the variates of VAR-GARCH model,
are these specifications not available to get the standard errors?

Any reply will be appreciated.

Owen Wang

I'm not even sure what "GIRF" means in that context. If you are talking about responses to unit shocks (which you *can* compute), the standard errors for those are can be computed using Gibbs sampling on the GARCH model---it the model is fairly well estimated, that can be done by independence chain Metropolis.