Performing parameter calibration numerically with Matlab, the solver needs to loop through parameter values and get theoretical moments. The computation time stoch_simul()
takes is around 1.5 to 2 seconds, thus when the solver is trying parameter values the program is very slow (for finding a calibration solution I’ll need probably tens of thousands of iterations). I’m solving my model at first order of accuracy with HP-filter 1600, theoretical moments computation, no periods of simulation, no irfs, is there a way I can reduce significantly the computation of stochastic simulation? Thanks!
Without seeing the files, it is impossible to tell.