Here's the Sims(2002) paper. It's equal to the published version.

The author gives the following example:

enter image description here

And tries to write it in the matrix notation of (1) in the paper.

However, his solution - in (4) - puts $\epsilon (t)$ in the first equation of (2) instead of $\nu (t-1)$, and $\nu (t-1)$ instead of $\epsilon (t)$. Is this correct, and why?

Also, when I try to write the the first equation of (2) the time notation doesn't match. The other two equations in (2) do match.

Any help would be appreciated.


1 Answer 1


Yes the $z(t)$ column vector, as it appears in $(4)$ is written upside-down in the translation. Also the first equation in $(2)$ is translated lagged once, i.e. the translated system describes the $w(t-1)$ equation.

Finally there is a typo in the $\Gamma_1$ matrix, there is a void 6th column that should be ignored, and $\theta$ should be in the 4th column not in the 5th (because it multiplies $u(t-1)$.

  • $\begingroup$ +1 Alecos, thanks for the help. By the way, in this paper, I don't understand how the author proposes to deal with the term $E_t W(t+2)$ (or $E_{t-1}W(t+1)$ when translated lagged once). In page 3, he suggests a solution, but I don't get the english. Why won't he just add $E_t W(t+2)$ to $y(t)$ as he did with $E_t W(t+1)$? $\endgroup$ May 23, 2018 at 14:31
  • $\begingroup$ Also, I think the C vector, should also have its 1st component multiplied by -1, since u_n and u(t) show up with different signs. $\endgroup$ May 23, 2018 at 14:33
  • $\begingroup$ You are right for the -1. $\endgroup$ May 23, 2018 at 19:05

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.