IRF with VARMA GARCH

[superleo]

Dear All,

My first attempt at modeling VARMA GARCH models with Eviews is turning ugly. In particular, I need to get Impulse Response Functions with bootstrapped confidence intervals.

As I do not see anything in Eviews for this, I am thinking about a two-stage process. First, to bootstrap/simulate the VARMA process, then to do the same with residuals, and then to add up the lower and upper bounds to get the right simulated confidence intervals.

That doesn't look stupid as everything is linear, but somehow that does feel ugly and dubious.

Any suggestion to get around this issue?

Thanks in advance,

Superleo