Bivariate EGARCH-M (1,1) Problems

[philphed]

Hello everybody,

I am fairly new to Eviews/Econometrics so I hope I do not bore you.
I am writing my msc dissertation on correlation of stock indices and I am working with a model from an already published paper (attachment).

I am really hoping that someone could help me.

Thank you!

[philphed]

Hello again,

after spending some more time on the problem and understanding a little bit more about Eviews, I got the model done. However now most of the coefficients from my conditional mean equation are not significantly different from zero..And I don't know where the problem is.
I will outline very quickly what I did and maybe someone knows where the mistake is:

- just to explain the problem again, I want to compare the correlation of returns of two countries and one part of that is to employ an EGARCH-M to do that
- the mean equation should look as follows: r(t) = c + a*Var(t-1) + b*Cov(t-1) + u(t) where r(t) is the return of e.g. Austria, C is just a constant, Var is the conditional variance and Cov is the conditional covariance of Austria and Germany
- to calculate the conditional covariance, I run two seperate EGARCH models with the simple mean equation r(t) = c + u(t) to obtain the conditional variances, then I calculated cov_germany_austria = @cor(germany, austria) * @sqrt(conditional_variance_germany) * @sqrt(conditional_variance_austria) where the correlation is just the correlation of the two return series
- when I had that, I was able to run the final model EGARCH-M with r(t) = c + a*Var(t-1) + b*Cov(t-1) + u(t) so using the Estimation input wizard, my mean equation looked like: 'austria c cov_germany_austria(-1)' and I also included 'in-mean' 'variance'
- I ignored the fact that Eviews conditional variance looks slightly different than Nelson's specification and will mainly concentrate on the coefficients of the mean equation. but as said before, most of them are not significantly different from zero, which is very different from the results in the paper I attached earlier
- I just thought, that since in EGARCH models, the conditional variance is log(sd^2), when I save the GARCH variance series, are the sd^2 saved or log(sd^2) saved?

Thank you and I would appreciate if someone could help me.


ps find one workfile attached, where couuntry_egarch is the simple egarch model where I obtained the conditional variance from to calculate the covariance and country_egarchm is the final EGARCH-M model