Perfect foresight simulation: only works under linear approximation

Hi there,
I’m trying to solve a perfect foresight model for a climate change transition path on some economic and environmental variables. The issue is that “standard” setup of perfect foresight (perfect_foresight_setup(periods=@{simulation_periods});
perfect_foresight_solver;)

does not work but, if I use the option “linear approximation” as perfect_foresight_setup(periods=@{simulation_periods});
perfect_foresight_solver(linear_approximation);

the model does indeed converge. I can’t explain myself why is this happening and if I can use the model solved in linear form anyways and, if so, how should I interpret the simulated paths of my variables (?).

I attach the code here in case someone can give me a hand on this.
Best,

J
sbonds_new.mod (12.3 KB)

Hi! Can someone help me with this?

This is a challenging issue. From what I can see, the simulation periods are insufficient to assure convergence to the new BGP. But increasing the periods is numerically problematic due to the exponential growth in your model.

Thanks for the reply!
I don’t really know how to make the model work without the need of a linear approx. In that regard, do you think it is too harmful to compute the solution in that manner? My guess is that the output should be quite similar without the linear approx

You need to judge that. You built the model and should have an intuition how nonlinear it is.