About the hp filter


the direction is : stoch_simul(order=1,hp_filter=100, periods=2100);

however matlab give me it as follows when i run dynare a.mod

ERROR: stoch_simul: HP filter is not yet implemented when computing empirical simulations

??? Error using ==> dynare at 96
DYNARE: preprocessing failed

how to deal with this problem?

I am duplicating one model economy with dynare, I got exactly the same policy function, however the standard error of each variable is much bigger than the results in the book. Will the filter matter?

why do we need the filters in dynare if the model economy is just fluctuating near the steady state?

another problem is that i found that we can not put periods=integ together with hp_filter=integ, can somebody tell me the reason?


  1. If you exclude the periods option, then Dynare provides theoretical moments of model variables for which the HP-filter can be applied. When the periods option is on, then the moments are simulated and the program does not have an option to yield HP-filtered data. (Although you could do this in a loop, by sending the simulated data to an HP-filter matlab code for repeated simulations)

  2. If the policy functions are the same, then you’re correct that the difference should be due to filtering.

  3. Whether you should filter a model variable which already yields stationary results is open to debate. I think the general practice is that you should conduct the same transformations to the model variables as you do to the data.