Re: [Gretl-devel] Granger causality
Allin Cottrell schrieb:
On Mon, 16 Nov 2009, Sven Schreiber wrote:
> I think gretl could use Granger causality tests (in a standard VAR
> framework). Actually I am going to upload a function package "real soon
> now" that does frequency-specific Granger causality tests (Breitung &
> Candelon 2006 Journal of Econometrics).
Sounds good! We do, however, have the basic Granger tests as part
of the default VAR output (i.e. F-tests on all lags of each of the
regressors).
ah, I missed that one because I was looking only quickly and was
watching out for the "Granger" keyword!
I would suggest to expand the explanatory string a little to mention
that it's in effect a Granger causality test.
For example:
"All lags of X (Granger non-caus.)"
If I have counted correctly, that should fit in one line together with
the formatted test output. Translators would have to be alerted to the
space constraints. And of course this does not apply when the own lags
of the respective endogenous variable in the equation are tested.
thanks,
sven