Page 1 of 1

Some theoritical queries related to BEKK GARCH

Posted: Sun Jun 07, 2020 9:45 am
by curiousresearcher
Dear All,

I have a couple of queries related to the BEKK-GARCH model

1) How to does it control for other factors e.g. when we are studying spillover between international crude oil prices and Chinese stock market returns, how does the BEKK-GARCH model control for other factors like currency etc.

2) If we are studying spillover between international crude oil prices and Indian stock market returns, where crude oil prices are in dollars
a) Should indian stock market returns be calculated in actutal stock index level denominated in Indian ruppe or dollar ? Will it matter ?

Kindly provide an explanation please . Any supporting academic paper will be very helpful so that i can read further

Thanks.

Re: Some theoritical queries related to BEKK GARCH

Posted: Thu Jun 11, 2020 8:08 pm
by TomDoan
vicky007 wrote:Dear All,

I have a couple of queries related to the BEKK-GARCH model

1) How to does it control for other factors e.g. when we are studying spillover between international crude oil prices and Chinese stock market returns, how does the BEKK-GARCH model control for other factors like currency etc.
BEKK-GARCH doesn't. Your choice of series does. A bivariate model with log oil prices in $ and log stocks (both in returns typically) and the same with log oil prices in yuan and log stocks will be different; which is "better" isn't necessarily clear.

On the other hand, a trivariate model with log oil in $, log stocks and log exchange rate will give an identical fit as the same with log oil in yuan, log stocks and log exchange rate since the three variables are connected by a simple linear transformation with a determinant of 1. Note, however, that while (for instance) the log likelihoods will be (effectively) identical and the time-varying variance of the stocks will as well, many of the BEKK coefficients will be different, as they apply to different variables. (If, for instance, the BEKK coefficients are diagonal in one representation, they will not and cannot be diagonal in the other).
vicky007 wrote: 2) If we are studying spillover between international crude oil prices and Indian stock market returns, where crude oil prices are in dollars
a) Should indian stock market returns be calculated in actutal stock index level denominated in Indian ruppe or dollar ? Will it matter ?
Again, it matters in the bivariate model and not (as far as fit goes) in a trivariate that includes the exchange rate.