I have a question related to the tolf option applied within the perfect foresight solver. It seems that I can only run my simulation (which involves a strong permanent policy shock) if I set tolf=1e-2. Now my question is: how can I test whether this is appropriate. My first guess is to look at steady state values of the model. If they are a greater order of magnitude than 1e-2 the convergence criterion should be ok. Is that correct? If not, what are the alternatives?