Determinacy graphs

Stochastic simulations
1

Hello everyone,

I am working with two versions of my model (the non linear version and the linear version) because I try to understand some mechanisms. However, I face the following issue and I was wondering whether is it normal or not. When I run the determinacy graphs in dynare by looping over some parameters, even if the shape of the determinacy graphs look the same between the two models, I have differences in the values that allow determinate dynamics in both cases. For instance, my model will be indeterminate for values of delta higher than 0.5 in one model while it would be higher than 0.4 in the other model. I checked for my linear equations but they look right, so I was wondering wether the determinacy analysis and determinacy graphs could lead to different threshold values for some parameters or is it not possible ?
Thanks in advance for your reply.

Best regards,

Hugo

2

No, what you describe indicates that the linear and the nonlinear one approximated at first order are not identical. At least one of the two must contain a mistake.

3

Hello Professor,

Ok thanks for your answer. Just another question, could it not be due to the fact that I have some variables in log-deviation while some others are in absolute deviation ? (because their steady-state is zero)

Best regards,

Hugo

4

No, if you correctly implemented everything, the results must be identical.

5

Ok perfect, thanks Professor.