I’m doing research using the TANK model based on Iacoviello(2005) with the Zero Lower Bound (ZLB) constraint. I am extremely struggling with some technical problems with OBC.
The mod file runs smoothly in the regular Dynare 5.3 version when the ZLB is not imposed.
However, I encountered issues with the stochastic simulation when I imposed the ZLB constraint on the Taylor rule and ran the model with Dynare OBC. It runs successfully when the number of periods is short (around 300), but it fails to run at all when I increase the number of periods to 100,000. The error message suggests increasing the Timetoescapebound parameter, but even with a sufficiently large Timetoescapebound value, the stochastic simulation still fails to run(or the MATLAB freezes).
Could you help me with how to solve this problem and properly conduct stochastic simulation with sufficient long periods with proper ZLB hitting frequency?
Secondly, I recommend installing a commercial solver like Gurobi (free for academics).
Thirdly, you should carefully read the initial DynareOBC output generated where it will report on the existence and uniqueness properties of your model. It sounds like your model does not always have a solution. A response to the price level can help. Or interest rate smoothing. Or lower responses to growth rates of output, if you have it. See my theory paper at REStat!
I forgot to mention that I am already using Dynare 4.5.7 when using Dynare OBC and installed GUROBI optimizer following recommendations in github.
Since my model does not have any growth component, I will try to modify Taylor rule part of my model.
My Taylor rule has smoothing, inflation responsiveness, output responsiveness, and specified as level.
Should I increase the smoothing parameter or incorporate the price level targeting component?