Dear Alexander;

thank you very much. I will look at it with great interest. Also, I forward it to Gianni Lombardo, with whom I'm having a discussion along similar lines, and to my colleagues in the Dynare Team.

All the best,

Michel

Dear Michel,

My coauthor and I have derived a perturbation solution using the nonlinear version of the infinite MA as an alternative to the standard state-space policy function that I thought might interest you. I have taken the liberty of attaching it to this email.

In brief, we find that taking the nonlinear MA approach yields stationary solutions (i.e., endogenously 'prunes') for simulations, decomposes nonlinear impulse responses into contributing elements from the underlying kernels of different orders (nicely illustrating the uncertainty component of responses to volatility shocks), and relates several invertibility assumptions of state-space solutions to more interpretable assumptions on the first order fundamental equation.

We've programmed it all up as an add on to Dynare (just need to add one line after stoch_simul) and I've attached the routines here. Of particular interest perhaps is the third-order `pruning' algorithm.

We would, of course, be thankful for any comments or suggestions you might have and are, in any case, grateful for your time.

Best, Alexander

-- Humboldt-Universitaet zu Berlin Institute for Economic Theory II Spandauer Strasse 1 10178 Berlin Germany Tel.: +49-30-2093-5720 Fax: +49-30-2093-5696 E-Mail: alexander.meyer-gohde@wiwi.hu-berlin.de