The RATS Software Forum

I am using BEKK model to estimate volatility and shock spillovers between three exchange rate returns series.Though i am getting volatility spillovers among all the three, i find unidirectional shock spillovers among two exchange rate pairs.Are my results correct? Is it necessary to get the shock spillovers when there are volatility spillovers among two variables?

Thanks
Sanjeev

sanjeev wrote:I am using BEKK model to estimate volatility and shock spillovers between three exchange rate returns series.Though i am getting volatility spillovers among all the three, i find unidirectional shock spillovers among two exchange rate pairs.Are my results correct?

Is that possible? Sure.

sanjeev wrote: Is it necessary to get the shock spillovers when there are volatility spillovers among two variables?

Absolutely not. One of the whole points of the GARCH model is that you can have lack of correlation in the 1st moment (mean model) while having very strong correlation in the 2nd moment (variances).

hi!

Thanks for the reply.I want further clarification on my query.I am analyzing three kinds of spillovers using VAR-BEKK model- returns spillovers,shock spillovers and volatility spillovers.I think spillovers in first moment are returns spillovers ,spillovers in volatility through squares of residuals are shock spillovers (information or news spillovers)while spillovers in the second moment are volatility spillovers.In my results, i am getting bi-directional returns as well as volatility spillovers among all exchange rate return pairs but i am getting uni-directional shock spillovers among two exchange rate return pairs.Are my results correct? Is it necessary to get information spillovers when volatility spillovers are present among two variables?

Thanks,
Sanjeev

Are your results possible? Yes. It's your job to determine if they are "correct".

Hi Tom, I thank you for your help. I would be grateful if you could possibly guide me again.
Could you suggest me the code for Hong Li paper titled “International linkages of the Chinese stock exchanges: a multivariate GARCH analysis”, Applied Financial Economics, March 2007.
Regards
Sanjeev

That's behind a pay wall, but from what I can tell, that's just a VAR(1) asymmetric BEKK GARCH model. GARCHMV.RPF has an example of a VAR(1)-BEKK model, so you would just add the ASYMMETRIC option to that.

Dear Tom,
Thank you very much for your help. I have another question. I am using VAR BEKK model on daily spot and futures returns of an agricultural commodity. I tried the @mvqstat(lags=5), the result shows that my mean model is not adequate. Kindly help. Kindly find attached herewith the data and command.
In addition, I tried near var bekk model also. Then I find that spot returns equation is adequate and futures returns equation is not adequate. What should I do?
I would be grateful if you could possibly guide me again.
Regards
Sanjeev

I'm not sure there's anything you can do about that. The second series, in particular, has a couple of enormous back-to-back outliers at 892-893. Neither the @MVARCHTEST nor the @MVQSTAT is designed to be robust to monsters like those. If those exist for exogenous reasons, you might want to dummy them out. (If you dummy them out in the mean, then they won't contaminate the variances). If the GARCH model is correct, then, if they were "endogenous" events, the series should be rather volatile for a considerable period of time after that. If, in fact, the subsequent data is more like other parts of the data range, then dummying them would make sense.

Hi Tom, how hope you are fine. I intend to forecast agricultural futures and spot commodity prices using the artificial neutral network model: could you send me the code of the following paper

Prabhati Kumari Misra Kishor Goswami , (2015),"Predictability of sugar futures: evidence from the
Indian commodity market", Agricultural Finance Review, Vol. 75 Iss 4.
Regards
Sanjeev

Neural networks are covered in the User's Guide.