One of the eigenvalues is close to 0/0! Important!!!

[jcgonzales]

Hello everybody! I'm new in dynare and I need some help. As well as the title mentions I have this problem with my code. There is an old topic posted about this but I don't really see I could apply the same thing here.

My model is: "Closed open economy model" with money and with adjustment cost of capital. It is taken from the book: "The ABCs or RBCs" of McCandless Chapter 13.

I didn't get any dynare code from the internet, not even the book web page has it because they only work with Matlab. However I tried to do it by my own and after several error-and-fix-it paths finally I got this result:

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Error using print_info (line 52)
One of the eigenvalues is close to 0/0 (the absolute value of numerator and denominator is
smaller than 1e-6)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

As well as I said, the old post mentions that it means Blanchard Khan conditions are not satisfied.

if this is the case my professeur from my univertisty told me once that when this occurs you should do something to fix it but he gave no clue what to do next.

I want to say something important: THE STEADY STATE RESULTS THAT I GET ARE ALMOST THE SAME AS THE BOOK HAS. So I don't think there is a problem with the SS.
Another thing I have already done is to change capital timing as well as dynare ask so there is no problem about that.
All the variables and shocks have their tag so you will know which they are

I need help with this please! So anyone that could help clarify this and get a solution will be welcome! I would really apreciate him/her.

my openeconomy.mod file is attached.

openeconomy.mod

This is my model done in dynare! the model is from McCandless chapter 13!

(5.36 KiB) Downloaded 217 times

[jcgonzales]

I have checked my problema and it seems it doesn't have relationship with Blanchard Khan condition. Actually, it is another problema. Could anyone tell me how to solve it?

Thanks!

[StephaneAdjemian]

Hi,

I didn't play with your mod file but I fixed something related to this error message yesterday.

A generalized eigengenvalue of 0/0 is related to Blanchard and Kahn because it means that any complex number can be a generalized
eigenvalue... So we don't have a unique solution in this case (the uniqueness problem here is different from the one treated by Blanchard
and Kahn which assume that the eigenvalues are uniquely determined and disscuss the uniqueness and solutions of the paths for the endogenous
variables). Obviously, in practice, we need to decide what is zero. By default, Dynare issues the 0/0 error message if the denominator and
numerator are less than 1e-6. Yesterday I added an option for this threshold level (called qz_zero_threshold). This option can be used in
check, stoch_simul and estimation commands. You can find an example on this gist.

If you believe that there is no problem with your model you may try to decrease the threshold level. The change is available in the snapshot version
of dynare and will will be included in dynare 4.4.0 (to be released next week).

Best,
Stéphane.

[selima]

Good morning dear Stephane Adjeman I tried to used this option check(qz_zero_threshold=1e-7); but I have obtained an error ERROR: essaifiscal.mod:878.7-23: syntax error, unexpected NAME.May be It works for a new version of dynare . I use 4.3.3 version

[jpfeifer]

As Stephane said, it is only available starting in Dynare 4.4