Dear Forum members,
I am trying to run a second order stochastic simulation using a previously estimated model.
As far as I understand, stochastic simulations of both orders should work, but I get some terrible “explosive” IRFs.
On the other hand, IRFs for the first order stochastic simulations look ok.
it is perfectly possible for a second order simulation to be explosive. There is no condition I know off that permits to tell in advance.
The only solution would be to use Chris Sims’ pruning algorithm, but it isn’t implemented yet in Dynare