Mixed shocks: Deterministic shock has no impact??

[Mark57]

Hi,
I am trying to run simulations using both stochastic and deterministic shocks. I started with the deterministic model (case of temporal shock) as in "Dynare User Guide" by Grifolli (Januare 2013; page 32). I found that k (capital stock) increases from 8 to 8.7 then converges back to 8. Then I added an exogenous shock (deterministic shock is included then using VAREXO_DET command; instead of SIMUL(PERIODS=2100); used STOCH_SIMUL(PERIODS=2100, IRF=0); FORECAST;) and found that deterministic shock has no impact on simulated time series (I changed the magnitude of the deterministic shock (from 0.1 to 0.0001) but results were the same). Further, k has never reached 8.7 value or even close (max value was 8.18). My best understanding was that in case of mixed shocks, the values of k would deviate along the values achieved using only deterministic shock. But it did not happen.
I was wondering whether DYNARE is not able to simulate models with mixed shocks or whether I failed to find where DYNARE store the simulated time series (I used oo_.endo_simul), or whether I did something wrong.
Thanks!
Mark

P.S. I am using Dynare 4.4.3 on Matlab R2015a.

[jpfeifer]

Please try the unstable version and report back.

[Mark57]

Thanks Johannes for prompt response.
The unstable version seems to be an improvement but I do not think it simulates the model with mixed shocks correctly (it is also possible that I do not know where to look for results).
Now, the dynamic of oo_.forecast.Mean.k (mixed shocks) is very similar to the dynamic of k (deterministic shock only), that is, they both are smooth upward sloping and then downward sloping lines (sim. to inverse v shape). However, oo_.forecast.Mean.k (mixed shocks) features no volatility that should be expected due to stochastic shock (its shape does not depend on the magnitude of the stochastic shock). Further, the behavior of k (mixed shocks; can be also found in oo_.endo_simul matrix) is not impacted by the magnitude of the deterministic shock (I tried 0.1 and 0.001 values of the temporal deterministic shock).
Regards,
Mark

[jpfeifer]

Maybe this file is helpful to get an insight into the workings: https://github.com/DynareTeam/dynare/blob/master/tests/forecast/Hansen_exo_det_forecast.mod

[Mark57]

Maybe this file is helpful to get an insight into the workings: https://github.com/DynareTeam/dynare/blob/master/tests/forecast/Hansen_exo_det_forecast.mod

Thanks Johannes,
Mark