[Gretl-users] gretl + dynare

Hello all, I've recently been messing with dynare, and noticed that since we support octave via gretl's "foreign" apparatus, we basically get dynare support for free. In light of that I've added support for dynare ".mod" files: you can open a .mod file in gretl's script editor, and send it for execution via octave by clicking the gear-wheel ("Run") button. (I should point out: If you have happen to have matlab installed, you should be able to get matlab + dynare support in gretl by going to /Tools/Preferences/General/Programs and entering the appropriate matlab path, or executable name, in the box titled "Path to octave executable".) Anyway, the ability to execute a dynare .mod file via octave or matlab from gretl's script editor is really just for the sake of completeness. For anyone interested in substantive interaction between gretl and dynare the more interesting workflow is likely to be of this sort: * Write a hansl script that calls octave/matlab using the "foreign" apparatus. Within the foreign block you can have octave/matlab call dynare and store results that you'd like to share with gretl. * Still within the gretl "foreign" block, make results available to gretl by using the gretl_export() function. * Back in gretl, read results from octave/matlab/dynare and analyse them or compare them with natively generated results. In case anyone is interested, I've put a couple of relevant files in http://users.wfu.edu/cottrell/ramsey/ * gr_dynare.inp : gretl/hansl driver script * gr_ramsey.mod : dynare model file called by the above You can open both of these in current gretl. Running the first may be of interest. I'm not going to try to explain the motivation, it would take too long, but if you're familiar with the Ramsey growth model it might be apparent. Brief word to the wise: I'd like to know, if you're heading for the Golden Rule steady state "from below", is there any advantage in respecting the Euler equation as opposed to just saving at the steady-state Golden Rule rate from the start (given a CRRA utility function with parameter sigma = 2): the answer appears to be Yes. Allin