Stochastic vs Deterministic simulation


I am working with a two-good open-economy model, with tradable and non-tradable sectors. I have perfect labour and capital mobility across sectors, which translates in equality of the rates of return.

Dynare finds the steady state without any problem, and Blanchard-Kahn are satisfied.

The models works fine when I implement stochastic simulations, of both order 1 and 2. When instead I try deterministic simulations, Dynare returns me an error message “Matrix is singular to working precisions”.

Did any of you experienced such an odd thing?


Based on my knowledge. The IRFs based on deterministic simulation are “calculated”, but the ones based on stochastic simulation are numerically approximated. The language is not accurate but the idea is the former one is more accurate.