8:36 p.m.
Sorry to bother you. I am a student from Nanjing Agriculture University China.
For my dissertation purpose I need to run Granger Causality tests using Gretl. I am using
time series data, would you mind giving me some advice on the following questions:
For everybody knows, Granger test is conducted in the framework of a vector autoregression
(VAR).But Granger causality tests are not provided among the menu in gretl when I
estimated a VAR, so how to run Granger Causality in Gretl?
1:57 a.m.
On Thu, 3 Mar 2016, ?????? wrote:
Sorry to bother you. I am a student from Nanjing Agriculture
University China.
For my dissertation purpose I need to run Granger Causality tests using
Gretl. I am using time series data, would you mind giving me some advice
on the following questions:
For everybody knows, Granger test is conducted in the framework of a
vector autoregression (VAR).But Granger causality tests are not provided
among the menu in gretl when I estimated a VAR, so how to run Granger
Causality in Gretl?
GC tests or each variable are reported under the corresponding OLS
regression; for example, if you run the code
<hansl>
open sw_ch14.gdt --quiet
series infl = 100*ldiff(PUNEW)
var 3 LHUR infl
</hansl>
you'll see the 4 GC tests as
<output>
All lags of LHUR F(3, 153) = 1668.1 [0.0000]
All lags of infl F(3, 153) = 9.5329 [0.0000]
...
All lags of LHUR F(3, 153) = 15.315 [0.0000]
All lags of infl F(3, 153) = 220.96 [0.0000]
</output>
HTH,
-------------------------------------------------------
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
-------------------------------------------------------
3:47 a.m.
Am 03.03.2016 um 07:57 schrieb Riccardo (Jack) Lucchetti:
GC tests or each variable are reported under the corresponding OLS
regression; for example, if you run the code
<hansl>
open sw_ch14.gdt --quiet
series infl = 100*ldiff(PUNEW)
var 3 LHUR infl
</hansl>
you'll see the 4 GC tests as
<output>
All lags of LHUR F(3, 153) = 1668.1 [0.0000]
All lags of infl F(3, 153) = 9.5329 [0.0000]
...
All lags of LHUR F(3, 153) = 15.315 [0.0000]
All lags of infl F(3, 153) = 220.96 [0.0000]
</output>
Of course, just 2 of the 4 are really Granger causality tests...
Actually it's not obvious why you want to test the exclusion of lagged
x-es in the x equation. Perhaps that line could even be removed from the
output, and then the explicit label of "Granger causality" could be
introduced.
cheers,
sven
5:26 a.m.
On Thu, 3 Mar 2016, Sven Schreiber wrote:
Am 03.03.2016 um 07:57 schrieb Riccardo (Jack) Lucchetti:
>
> GC tests or each variable are reported under the corresponding OLS
> regression; for example, if you run the code
>
> <hansl>
> open sw_ch14.gdt --quiet
> series infl = 100*ldiff(PUNEW)
> var 3 LHUR infl
> </hansl>
>
> you'll see the 4 GC tests as
>
> <output>
> All lags of LHUR F(3, 153) = 1668.1 [0.0000]
> All lags of infl F(3, 153) = 9.5329 [0.0000]
>
> ...
>
> All lags of LHUR F(3, 153) = 15.315 [0.0000]
> All lags of infl F(3, 153) = 220.96 [0.0000]
> </output>
>
Of course, just 2 of the 4 are really Granger causality tests...
At the risk of being pedantic: it's perfectly possible for a variable to
be Granger-noncausal for itself, so I would say those are 4 legit GC
tests.
-------------------------------------------------------
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
-------------------------------------------------------
6:03 a.m.
Am 03.03.2016 um 11:26 schrieb Riccardo (Jack) Lucchetti:
On Thu, 3 Mar 2016, Sven Schreiber wrote:
> Of course, just 2 of the 4 are really Granger causality tests...
At the risk of being pedantic: it's perfectly possible for a variable to
be Granger-noncausal for itself, so I would say those are 4 legit GC tests.
I have never seen or heard anyone advocate such a definition, so please
give references.
Typically I'd say the need for lagged values of the LHS variable is
called persistence (or anti-persistence), not "own causality".
thanks,
sven
6:17 a.m.
On Thu, 3 Mar 2016, Sven Schreiber wrote:
Am 03.03.2016 um 11:26 schrieb Riccardo (Jack) Lucchetti:
> On Thu, 3 Mar 2016, Sven Schreiber wrote:
>
>> Of course, just 2 of the 4 are really Granger causality tests...
>
> At the risk of being pedantic: it's perfectly possible for a variable to
> be Granger-noncausal for itself, so I would say those are 4 legit GC tests.
>
I have never seen or heard anyone advocate such a definition, so please give
references.
You heard me :)
If you take the standard definition of GC from x to y, I don't see why one
should implicitly rule out the case x = y.
Typically I'd say the need for lagged values of the LHS variable
is called
persistence (or anti-persistence), not "own causality".
A pet peeve of mine has been (for many years), that what we call Granger
Causality is neither Granger's, nor causality. It should be called
"predictive power". But maybe it's just me.
-------------------------------------------------------
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
-------------------------------------------------------
11:46 a.m.
Am 03.03.2016 um 12:17 schrieb Riccardo (Jack) Lucchetti:
On Thu, 3 Mar 2016, Sven Schreiber wrote:
> I have never seen or heard anyone advocate such a definition, so
> please give references.
You heard me :)
If you take the standard definition of GC from x to y, I don't see why
one should implicitly rule out the case x = y.
Well until you give another reference, you can call it "Lucchetti
causality", but not Granger's!
> Typically I'd say the need for lagged values of the LHS variable is
> called persistence (or anti-persistence), not "own causality".
A pet peeve of mine has been (for many years), that what we call Granger
Causality is neither Granger's, nor causality. It should be called
"predictive power". But maybe it's just me.
Well everybody agrees that GC is about predictive power. There's a lot
that could be said about the rest, but that would mean to repeat
discussions that have been settled (or left unsettled) before (not
between us, not on this list).
cheers,
sven
3118
days inactive
3118
days old
6 comments
3 participants
participants (3)
-
Riccardo (Jack) Lucchetti -
Sven Schreiber -
何晨蕾