Markov Regime Switching GARCH for VaR

njindian

njindian

Joined
2/9/13
Messages
18
Points
13
Hi,

I need a guidance on AR1, GARCH1,1 modelling. My data is from 2004-2013 so includes the volatility in subprime period. The data can be divided into three periods – 2004-08,2009-11,2012-14 based on returns trends for the 5 stocks. So, I think we shouldn't do volatility modelling for VaR without accounting for crisis period and account for regime change use something like a Markov Regime Switching GARCH (MRS-GARCH) model.

Any guidance would be welcome.

Thanks,
Nitin
 
I would probably help if I knew exactly what you need guidance with.
look up HAMILTON 1989 paper on regime switching time series. He pioneered that idea and provided a decent filter for estimation of switching time series. I haven't worked on switching Garch yet but I bet the model has been worked out already. find the right paper rewrite the code and get the estimates. Better yet there should be a Matlab code available that would probably help.
 
Hi, Thanks for the reply. I have returns of 5 stocks from 2003-2014. Need to build and AR1, GARCH1,1 model to compute VaR. Trends show that there are three sub periods - 2003-08,2009-11,2012-14. I think due to such volatile periods I shouldn't use the whole data but should split and then do the analysis and later weight the volatilities. Or use something like MRS-GARCH. Pls suggest something that can be used for such a data.
 
Hey njindian, I have done a similiar topic using the MSVAR model. But for MS-GARCH I am not that familiar. Can I ask what tools you used for doing the model? Do you write the code by yourself?
 
Back
Top Bottom