I am trying to embody stochastic volatility in a small scale-NK model according to method 3 shown in Andreasen (2010) paper: https://www.sciencedirect.com/science/article/pii/S0165176510000923
These are the features I have added to a standard model definition:
In the VAR block:
A Technological process - final
V Technological shock process
sigA Technological volatility
rho_V Persistence of a technology shock
rhosigA Persistence of the standard deviation shock
e_V Technology shock
e_sigA Standard deviation shock
In the model block:
[name=‘12. Technology shock process’]
log(A) = sigA*log(V);
[name=‘13. Technology shock - v’]
log(V) = rho_V*(sigA(-1)/sigA)*log(V(-1)) + e_V ;
[name=‘14. Shock to the standard deviation - sigA’]
log(sigA) = rhosigA*log(sigA(-1)) + e_sigA ;
The attached mod file seems to work.
Do you think what I am doing makes sense?
I also attach the IRFs ralated to the technology and volatility shock.
Is it normal that the latter generates non-smooth IRF?
Is there sample code in one of the dynare repository?
Many thanks in advance.
Any input would be very appreciated.