is this right process to get the granger causality and impulse response after the Johansen cointegration?
Posted: Wed Apr 25, 2018 3:30 am
Dear Folks,
This is a quite a long process to estimate the granger causality and impuse response after the Johansen cointegration results.
My questions is this one;
When we found one cointegrating relation between the data group, the next step is to keeping the cointegrating relation and produce the results of impulse response and granger causality. I feel that the process explained in the eviews manuals would not be clear to my poor understanding.
So, after finding the one or two cointegrating relation,
1) then we go to open "estimation VAR" and
2) then here comes VAR specification and choose Vector error correction to fill out the endogenous and lag
interval. Also, fill out the specification of cointegration window.
3) well prepared, than run the VECM model under cointegration relation.
4) Here we have Vector Error Correction Estimates.
my question is from here;
if i click the button of impulse response in the window at this point, does this imply that the results of impulse response are the results produced from the cointegrating relationship which was already found by Johansen estimation.
Also, if I click the lag structure in the view and select "granger causality/block exogeniety", does this results of granger causality also would be produced from the same cointegrating relation as like the above impulse response?
The manual is not precisely guiding me; so, I am not sure of what i did follow this way is correct ones at Eviews.
Your kind comments are really helpful and great!
Thank you very much!
Insang
This is a quite a long process to estimate the granger causality and impuse response after the Johansen cointegration results.
My questions is this one;
When we found one cointegrating relation between the data group, the next step is to keeping the cointegrating relation and produce the results of impulse response and granger causality. I feel that the process explained in the eviews manuals would not be clear to my poor understanding.
So, after finding the one or two cointegrating relation,
1) then we go to open "estimation VAR" and
2) then here comes VAR specification and choose Vector error correction to fill out the endogenous and lag
interval. Also, fill out the specification of cointegration window.
3) well prepared, than run the VECM model under cointegration relation.
4) Here we have Vector Error Correction Estimates.
my question is from here;
if i click the button of impulse response in the window at this point, does this imply that the results of impulse response are the results produced from the cointegrating relationship which was already found by Johansen estimation.
Also, if I click the lag structure in the view and select "granger causality/block exogeniety", does this results of granger causality also would be produced from the same cointegrating relation as like the above impulse response?
The manual is not precisely guiding me; so, I am not sure of what i did follow this way is correct ones at Eviews.
Your kind comments are really helpful and great!
Thank you very much!
Insang