Re: [Gretl-devel] Gretl VAR Improvements
On Mon, 29 Mar 2010, Riccardo (Jack) Lucchetti wrote:
On Thu, 25 Mar 2010, Henrique Andrade wrote:
> I would like to suggest some improvements (I think ;-)) to the
> VAR model window...
>
> (1) Impulse-response: I think it could be useful if we had
> more options regarding to the impulses. EViews has the option
> "impulse definition" where we can choose the decomposition
> method (Residual - one unit; residual - one std. deviation;
> Cholesky - dof adjusted; Cholesky - no dof adjustment;
> generalized impulses; structural decomposition; and user
> specified.
[...]
Cholesky: dof/no-dof. Do you reaaly think this would be of any use?
Like Jack, I'm having difficulty seeing why offering a choice of
the size of the innovation (MLE sigma-hat, df-adjusted s, unit
shock, whatever) for an impulse response plot should be worth
doing.
Since a VAR is a linear system, this just amounts to printing
different numbers on the y-axis, while the profile of the response
is identical. I'd have thought it was part of the mental toolkit
of an economist to be able to rescale such things if needed, much
as we manage to figure t-ratios to a reasonable approximation when
we're given just a coefficient and its standard error. And since
VAR impulse responses almost always have a wide band of
uncertainty, a reasonable approximation is truly all that's
required.
Allin.