VAR-GARCH

For questions regarding programming in the EViews programming language.

Moderators: EViews Gareth, EViews Jason, EViews Moderator, EViews Matt

mollydokora
Posts: 1
Joined: Tue Sep 23, 2014 2:10 am

VAR-GARCH

Postby mollydokora » Wed Oct 01, 2014 1:09 pm

Hi everybody,

I'm currently working on my thesis and I should estimate the temporal volatility transmission across the crude oil market and different stock market sectors . I have the time series for oil and stock market indices logged from 1995 to 2013. I have EVIEWS 8.

I would like to ask if there is a way to build a bivariate VAR(1)-GARCH(1,1) model on EVIEWS: seems that the best software is RATS, but I have no idea how to use it.

I am replicating a study by Arouri, Jouini and Khuong 2012 and have no idea how to program this model in eviews.
Does anybody have an idea?

Thank you so much!

bvguizar
Posts: 40
Joined: Sat Aug 28, 2010 6:56 am

Re: VAR-GARCH

Postby bvguizar » Fri Dec 19, 2014 5:10 am

Hi,

Did you get a reply about this? I am trying to do the same thing but for ethanol market.

trubador
Did you use forum search?
Posts: 1518
Joined: Thu Nov 20, 2008 12:04 pm

Re: VAR-GARCH

Postby trubador » Fri Dec 19, 2014 6:30 am


bvguizar
Posts: 40
Joined: Sat Aug 28, 2010 6:56 am

Re: VAR-GARCH

Postby bvguizar » Tue Dec 30, 2014 8:28 am

Dear Trubador,

How can you calculate spillover ratios using BEKK or DCC models?

trubador
Did you use forum search?
Posts: 1518
Joined: Thu Nov 20, 2008 12:04 pm

Re: VAR-GARCH

Postby trubador » Wed Dec 31, 2014 2:42 am


bvguizar
Posts: 40
Joined: Sat Aug 28, 2010 6:56 am

Re: VAR-GARCH

Postby bvguizar » Fri Jan 02, 2015 9:30 am

Dear Trubador,

I went through the post you sent and I was wondering if the programming for spillover ratios can be done using three variables?

I have done the programming again but when I ran it I see many problems. Please find attached the program and the E-views file. Could you help me with this?
Attachments
eviews_example garch model.WF1
Eviews File
(21.6 KiB) Downloaded 538 times
BV Garch in mean_Lowertriangle EVIEWS EXAMPLE for 3 variables.docx
Programming
(20.28 KiB) Downloaded 521 times

trubador
Did you use forum search?
Posts: 1518
Joined: Thu Nov 20, 2008 12:04 pm

Re: VAR-GARCH

Postby trubador » Mon Jan 05, 2015 1:23 am

That code estimates a bivariate garch-in-mean model as the name suggests. You need to get rid of in-mean specifications and modify the rest of the code so as to represent a VAR-GARCH model. It may be tedious, but not difficult. Even if you establish the correct specification, you may still experience some convergence problems. In that case, please search the forum for similar discussions.

bvguizar
Posts: 40
Joined: Sat Aug 28, 2010 6:56 am

Re: VAR-GARCH

Postby bvguizar » Mon Jan 05, 2015 5:49 am

Dear Trubador,

I am new to programming. I'm not sure where to put a VAR-GARCH model in here?

However, I have checked other posts

viewtopic.php?f=4&t=1298&p=26964#p26964

viewtopic.php?f=4&t=9476

I thought I have changed the bivariate model to put it in 3 variables. So, a mean equation is not needed in a VAR-GARCH model?

Could you help me please?

trubador
Did you use forum search?
Posts: 1518
Joined: Thu Nov 20, 2008 12:04 pm

Re: VAR-GARCH

Postby trubador » Mon Jan 05, 2015 7:00 am

bvguizar wrote:I am new to programming. I'm not sure where to put a VAR-GARCH model in here?

The issue here appears to be more demanding than your programming skills. Your task requires technical background regarding: i) Spillover effects, ii) VAR-GARCH models and iii) Maximum likelihood estimation.
bvguizar wrote:I thought I have changed the bivariate model to put it in 3 variables. So, a mean equation is not needed in a VAR-GARCH model?

You confuse "in-mean" concept with "mean equation". The first one refers to inclusion of conditional volatility in the mean equation. In your case, mean equations should look like a VAR(1) model and the MGARCH part should operate on the residuals from those equations. And since you are interested in spillover effects, diagonal specifications will not work for you.

bvguizar
Posts: 40
Joined: Sat Aug 28, 2010 6:56 am

Re: VAR-GARCH

Postby bvguizar » Mon Jan 05, 2015 7:06 am

trubador wrote:
bvguizar wrote:I am new to programming. I'm not sure where to put a VAR-GARCH model in here?

The issue here appears to be more demanding than your programming skills. Your task requires technical background regarding: i) Spillover effects, ii) VAR-GARCH models and iii) Maximum likelihood estimation.
bvguizar wrote:I thought I have changed the bivariate model to put it in 3 variables. So, a mean equation is not needed in a VAR-GARCH model?

You confuse "in-mean" concept with "mean equation". The first one refers to inclusion of conditional volatility in the mean equation. In your case, mean equations should look like a VAR(1) model and the MGARCH part should operate on the residuals from those equations. And since you are interested in spillover effects, diagonal specifications will not work for you.


I am sorry for the trouble, yes I do have issue with this since I just got started to familiarise in these GARCH models. I have calculated a VAR in Microfit and I am using the residuals of the VAR to calculate the spillover effects. I wasn't aware of the limitations using diagonal specifications. In theory I assumed that once you estimate the BEKK model, you calculate the conditional variance of steady state. In a BEKK model with 3 series, for example, you are estimating 6 equations (3 variances and 3 covariances). Then you assume that Ht=Ht-1 and episilon t-1=0 (you are in equilibrium). This is how you find h11, h22, h33, h12, h13, h23 of steady state (six equations & six unknowns). And finally you simulate a shock in one market that changes the conditional variance (of steady state) by 1% and see how this affects the variances in all markets over a period of time. Can this be done in Eviews?

trubador
Did you use forum search?
Posts: 1518
Joined: Thu Nov 20, 2008 12:04 pm

Re: VAR-GARCH

Postby trubador » Tue Jan 06, 2015 2:53 am

It is doable, but you may need to put a lot more effort into it than you think.


Return to “Programming”

Who is online

Users browsing this forum: No registered users and 26 guests