var-diagonal bekk model

[sanjeev]

Hi, i am using two time series variables, futures and spot price. I want the code for var- diagonal bekk model. I have used the code given by you http://www.estima.com/forum/viewtopic.php?f=11&t=495). What should be the code if mean equation is var in diagonal bekk model.
Thanks, Sanjeev

[TomDoan]

Note that MV=TBEKK (for triangular BEKK) and MV=DBEKK (for diagonal BEKK) were added to the GARCH instruction with RATS 8.3. Also, the SIGNS option was added to allow for different sign effects on asymmetry.

So just use a VAR as the mean model and MV=DBEKK on a GARCH instruction.

[sanjeev]

Thanks Tom, Is there a way to export H11, h22, h12 matrices in excel. I tried but in vain.

[TomDoan]

In what context? The program that you cited doesn't ever use those. With the GARCH instruction, you would use the MVHSERIES option and do a COPY instruction. If you're generating the covariance matrices some other way, you need to save the histories as separate series and then use COPY on those.

[sanjeev]

Thank you very much for your reply Tom. I got the solution.
I have another question about my model. I would be grateful if you could possibly guide me again.
Can I find the average of H11, h22, h12 in rats? Actually I am replicating the Chang et al 2011 paper titled “Crude oil hedging strategies using dynamic multivariate GARCH” I need to calculate average hedge ratio, average optimum weights.
Thanks, Sanjeev

[TomDoan]

Sadorsky, EE 2012 does hedge ratios. Your calculations should probably be similar.

[sanjeev]

Hi Tom, I am using the following code for estimation of VAR-bekk model: garch(p=1,q=1,model=var1,mv=bekk,pmethod=simplex,piters=10,hmatrices=hh,mvhseries=kk,rvectors=U).

But I am not getting the AIC and SIC values. How can I get AIC and SIC measure for this model. Thanks Sanjeev

[TomDoan]

Use @REGCRITS

[sanjeev]

Thank you very much for your reply Tom. I got the point.
I have another question about my model. I would be grateful if you could possibly guide me again. could you give me the code for calculating optimal lag in VAR model. AIC, SIC and HQ criterion gives different lags for the var model. which one should I take for the optimal lag for var model.
Thanks Sanjeev

[TomDoan]

First of all, are you making sure that you run the estimation over a common range when you change the lags? By default, it will take a longer range when the lags are shorter, which makes the IC's not comparable. For a VAR-GARCH model, I tend to prefer picking the smallest reasonable lag and then checking for residual autocorrelation. This all covered in the GARCH e-course.

[sanjeev]

Hi Tom, could you help me with the code for multivariate diagnostic test in BEKK/dcc/cc/dbekk/diag vech model. Thanks Sanjeev

[TomDoan]

Have you read the section in the User's Guide? There's no difference between the diagnostics for the different forms of models.

[sanjeev]

Hi Tom,
I am using six time series variables, i.e., futures returns of six agricultural crops. I am interested in finding the volatility spill over from one variable to other. I am using the following code for estimation of VAR-bekk model:
: *
system(model=var1)
variables dlbarfpsq dlwheatfpsq dlcorifpsq dlgramfpsq dlzeerafpsq dlmustfpsq
lags 1 2
det constant
end(system)
*
garch(model=var1,p=1,q=1,mv=bekk,method=bfgs,iters=500,$
pmethod=simplex,piters=10,robusterrors,rvectors=rbekk,hmatrices=hbekk,stdresids=zu)

However, there is a convergence problem. Could you please suggest me the code so that there is a convergence.
Regards,
Sanjeev

[TomDoan]

diagonal-VECH is just the default option (MV=STANDARD) for the GARCH instruction.

[ABDHUT123]

how to summarize the parameter of Diagonal BEKK model.......for example suppose in a bivariate case if I want to test ax(1,1)*a(2,2)=0 how to summarize this....