I have the following question/problem regarding dynare++.
I am using dynare++ to solve an optimal monetary policy problem in a currency union with 2 countries. I am mainly interested
in a 2nd order approximation (note: only a 1st order appr. is available in the normal dynare). My problem is that some of the results
that dynare++ produces do not seem to be comparable with the “normal” version of dynare and that some results are unstable.
If I start out with a closed economy (no union) and use the optimal policy command, then dynare++ (1st order and 2nd appr.)
produces the same results as dynare 4.3.0.
If I introduce a second completely identical country and form a currency union there is still no real difference between dynare++ and
dynare 4.3.0 under the 1st order appr. The results of the 2nd order appr. of dynare++ are also close to the 1st order appr.
However, if the 2 countries differ in terms of the parametrization, strange things happen. There are relatively large differences
(in terms of the variances of the variables and the IRF’s) between dynare++ and dynare 4.3.0 under a 1st order appr. Furthermore, the
results that dynare++ produces under a 2nd order appr. are very unreliable, that is, repeating the same simulation (or only changing
the number of periods of the simulation) can yield very different results.
Note that these problems do not occur when I use Taylor rules instead of the optimal policy command.
Has anyone encountered similar problems under optimal policy with dynare++ or some useful comments?