Homotopy for large deterministic shocks

Perfect foresight simulations
1

Hello,

I am running a model with large deterministic shocks and relying on the homotopy method in Dynare. Most of the time the method arrives at a solution, but in a few cases the method does not find a solution, even though lambda is very close to 1 (greater than 0.99). From what I understand this means that the solution shown is only marginally far from the “true” solution. Am I correct?

Thank you for your time and your extraordinary work to make modelers’ lives easier.

2

Yes, that should mean that you are very close to the solution to the true problem. The tricky part is understanding why the last part does not work, e.g. a solution may not exist.

3

Thank you for your response, Johannes. I can live with a shock which is an epsilon lower and for which a solution exists.