enter image description here

I know I have to check the eigenvalues for the stability condition in general but I do not know how to process this exercise. How can I bring it in the right form? How do you check c) ?

enter image description hereCan anyone help with showing the single steps?

For c) I would say that c does not G-cause y becaue of the 0 in the matrix. But y G-causes c.

  • 1
    $\begingroup$ Granger causality in the variable of some equation is present if any of the lags of the other variable(s - but here we only have one) are nonzero. $\endgroup$ Commented Jan 23, 2017 at 5:09
  • $\begingroup$ Thank you. How would you attempt a); I would check for unit roots. $\endgroup$ Commented Jan 23, 2017 at 14:12
  • $\begingroup$ @Christoph Hanck how can I brind the equation in world representation from? $\endgroup$ Commented Jan 23, 2017 at 21:37
  • $\begingroup$ for a) you indeed may look at the appropriate eigenvalues $\endgroup$ Commented Jan 24, 2017 at 5:51
  • $\begingroup$ How can I bring the given equation in the right format? $\endgroup$ Commented Jan 24, 2017 at 18:54

1 Answer 1


You need to check the roots of $$ det(I-Az)=0, $$ where $A$ is your coefficient matrix on the lags.

Thus, $$ det\left(I-\begin{pmatrix}\alpha&\beta\\0&\gamma\end{pmatrix}z\right)=det\begin{pmatrix}1-\alpha z&-\beta\\-0&1-\gamma z\end{pmatrix}=0 $$ or $$ (1-\alpha z)(1-\gamma z)=0 $$ which has solutions $$z_1=1/\alpha\quad z_2=1/\gamma,$$ which need to be outside the unit circle. You can take it from here.

  • $\begingroup$ For a) I would suggest that alpha and gamma need to be between 0 and 1 to have a stable process because then the roots are bigger than 1 in absolute value. $\endgroup$ Commented Jan 28, 2017 at 9:08
  • $\begingroup$ Yeah, I have the same result. But then I also defined the parameters and said that $\alpha$ has to be between 0 and 1 (and between -1 and 0) as I need z bigger than 1 in absolute values to define a stable system. $\endgroup$ Commented Jan 28, 2017 at 11:20
  • $\begingroup$ Right, that is what I'm saying. $\endgroup$ Commented Jan 28, 2017 at 11:24

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.