VAR with time-varying parameters and stochastic volatility

[linea]

Thanks for the reply, Todd. But I think we are still not on the same page.
These are the notes from the January 2014 version.
FIRST: 07-25-2010
* REVISED: 05-02-2013 TO CORRECT THE ORDERING OF GIBBS STEPS PER DEL NEGRO AND PRIMICERI (2013)
* REVISED: 01-27-2014 to correct an error in not initializing log volatility before MCMC loop

If I understand correctly, the January 2014 version corrects for Del Negro and Primiceri (2013), the original 2010 does not. And the program I attached in my previous post (here it is again) does not work with the January 2014 version. It works fine with the original 2010 version, but that seems not to be what I am interested here.

Please correct me if I'm wrong.

[tclark]

The 2010 code is not code you should use; it has the error of Primiceri's original code. You should use the 2014 version. As to the cause of the error you are now getting, Tom has pointed out some previous errors in your implementation, and I would encourage you to check your code for other possibilities.

[istiak]

Dear Team,
The identification here is based on recursive structure. Is there any way to impose other identification structures by using zero restrictions only? If yes, can you please help?
Thank you.

[ege_man]

Dear Tom
One of the reviewer is asked me to estimate TVP-VAR model using Minnesota prior as a robustness check. Is it possible to modify the code for this? Or if not can you refer me any paper to answer this response?
Thanks

[ege_man]

Dear Tom,
Can you refer me any paper for this expression used in the introduction of VAR e-course?
"Strongly informative priors (such as the so-called Minnesota prior) are widely used for building forecasting models, but they tend to improve forecasts by shutting down much of the cross-variable interaction. "
in order to response the reviewer request I will also try to estimate VAR model in the following link https://estima.com/forum/viewtopic.php?f=4&t=2886 and then compare them with one using Gibbs sampling. Do you think thank it is ok?
Best

[tclark]

If I might make a suggestion, good starting points would be the chapters "Bayesian Forecasting" by Geweke and Whiteman in vol 1 of the Handbook of Economic Forecasting and "Forecasting with Bayesian Vector Autoregression" by Karlsson in vol 2 of the Handbook of Economic Forecasting.

[ege_man]

Dear Tom,
Instead of selecting important dates is it possible to draw responses over the whole estimation sample? Can you show me how to modify it? I would like used it to show the impact of oil price on daily stock returns.
I previously used this code to get responses at each point of time by inputting all dates to the vector comp [vec[int]] impdates = on line 333.
Best Regards