non-linear dynamics with simul

[petros_varth]

Hi all,

I use a two country model. I study the deterministic transition from the old steady-state to a new steady-state. But i come up with the following problem:

When i run the simul command for 100 periods i get convergency obtained and the paths of the endogenous variables make sense.

But when i run it for 200 periods i get the following error :
Warning: Matrix is close to singular or badly scaled.
Results may be inaccurate. RCOND = 3.049764e-017.
> In sim1 at 129
In simul at 100
In two_country_welfare_sg at 659
In dynare at 120

and at the end Convergency obtained.

But the vectors of the endogenous variables contain Nan+Nani.

Does anybody have any idea why this happens?

Thanks in advance.

P.

[petros_varth]

Hi,

Please let me add the following. In the case with 100 periods some endogenous variables have explosive paths so this solution does not converge but why the algorithm converges? Are 100 periods too few probably?

P.

[jpfeifer]

Sorry, but the convergence message is wrong. This is a bug in 4.3.3, see http://www.dynare.org/DynareWiki/KnownBugs.

Have you tried model_diagnostics?