Conflicting results of unit root test and AR root graph
Posted: Thu Aug 06, 2015 10:16 am
Hello,
how does the results of the unit root test relate to the AR root graph?
To mention a specific example:
Before I am estimating my VAR model, I take every endogenous variable of my VAR model and test it for stationary using the ADF test.
For my model, I get the result that all variables are stationary.
Next, I estimate my VAR model and test again for stability using the AR root graph. All modulus are less than one, which suggests stability.
However, even if I include variables which are not stationary according to the ADF test, the AR root graph shows that all modulus are less than one, again.
How could this happen? I thought that if I include variables which are not stable, the AR root graph must show modulus greater than one, meaning that the system is not stable.
Thanks for any support!
how does the results of the unit root test relate to the AR root graph?
To mention a specific example:
Before I am estimating my VAR model, I take every endogenous variable of my VAR model and test it for stationary using the ADF test.
For my model, I get the result that all variables are stationary.
Next, I estimate my VAR model and test again for stability using the AR root graph. All modulus are less than one, which suggests stability.
However, even if I include variables which are not stationary according to the ADF test, the AR root graph shows that all modulus are less than one, again.
How could this happen? I thought that if I include variables which are not stable, the AR root graph must show modulus greater than one, meaning that the system is not stable.
Thanks for any support!