Re: [Gretl-devel] [Gretl-users] multivariate lrvar?
On Mon, 16 Apr 2018, Sven Schreiber wrote:
Am 16.04.2018 um 18:29 schrieb Riccardo (Jack) Lucchetti:
> On Mon, 16 Apr 2018, Sven Schreiber wrote:
>
>> Am 16.04.2018 um 18:07 schrieb Riccardo (Jack) Lucchetti:
>>> Hm, I see. I personally would prefer redefining the $\Gamma$ matrices.
>>
>> Fine with me. Only that the sentence "... expresses \Sigma as the long-run
>> covariance of..." would also have to be adapted with a phrase like
>> "proportional to" or something.
>
> I'd prefer "expresses $\Sigma$ as the long-run covariance matrix of the
> random $k$-vector $x_t \cdot u_t$", since $X$ is a matrix.
Right, that as well. But with the T-factor now Sigma won't be the long-run
cov matrix of that animal anymore, that was my other point.
Uhm, no. The present expression for $\Sigma$ would be right in that case,
since the quantity $V($x_t \cdot u_t)$ is independent of $T$, whereas
$E(X'uu'X)$ isn't.
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------