Problem in perfect foresight case. Works in stochastic

Perfect foresight simulations
1

Dear all,

The model attached is a Bernkanke-Gertler-Gilchrist '99 type model. Although it performs fine in the stochastic case with sensible results, convergence cannot be achieved in the deterministic case ith any size of shock > 0BGGFlexiDet_steadystate.m (3.9 KB)
myfun_BGG_calib.m (746 Bytes)
BGGFlexiDet.mod (6.0 KB)
. I have the following questions:

  1. If you could kindly explain my the source of the convergence problem.
  2. If I want to put as endval of the exogenous distrurbance a value different from its ss value (=0) how is that possible with an external ss file like mine? Since A (TFP) is always 1 in ss, how can I make it 1 + (the endvalue of the disturbance)?
2
  1. First of all, you need to provide correct initval and endval-blocks. Unless you put a steady after those blocks, all variables not specified at assumed to be 0. In a nonlinear model that cannot work.
  2. Your endval need not be a steady state. You can end in any condition you like. The only thing is that usually a steady state makes sense.