The RATS Software Forum

plz check table 3.

Yes. That looks like that will estimate the type of model that they're doing. (They don't report the CONSTANT in the mean equations, but the formula does include an intercept).

omid_ie wrote:Dear Tom;

I just really want to run VAR(1) with variance following : CCC- VARMA (Mcaleer 2003).

The thing is I dont have just 2 indices it is 31 pairs!!

I will do the test for residuals but as there 31 pairs of data, so still I need to check VAR(1)
so please help me to finish this. thanx a lot.

Again, running a VAR(1) will rarely cause any problems even if the lag terms aren't really necessary. It's the VARMA mean that basically won't work if there's no serial correlation to start.

31 pairs is reasonably straightforward using DO or DOFOR loops. However, 31 seems like a strange number. You're not doing all combinations of two?

Sir,
I hope you are well and sound. I have manage to get WinRats . I am trying conduct similar study like 'Modelling international tourism and country risk spillovers for Cyprus and Malta'(Tourism Management ,vol 28, issue 6 and page 1472-1484) by using VARMA-GARCH(2003) and VARMA-AGARCH(2009). I could not found QMLE in RATS and i am not sure if (robust (HAC) Standard Errors is for QMLE ? if I open rats and put data then go to TIME SERIES> ARCH/GARCH MULTIVARIATE> Dependent Variable(Tourist arrival for country x)>model=CC> Distribution=Varma for VARMA-GARCH and TICK in ASYMMETRIC EFFECTS for VARMA-AGARCH and tick on Robust standard errors then WILL I GET THE RESULT ?

sorry for asking too much !

Thank you vary much for your kind help .

Kind Regards

Rubel

That's correct. You estimate with Gaussian errors (the default) and add ROBUSTERRORS. For more, see

https://estima.com/forum/viewtopic.php?f=6&t=731

Sir,
I hope you are well and sound. I will glad if you could please help me to estimate spillover effect parameters. I have four variable( Country risk return for UK, Tourism growth rate of UK, Country risk return of France and Tourism Growth rate of France) like the paper of Hoti & McAleer 2007, vol-28,issue-6,page 1472-1484 where they have used Malta and Cyprus . If you please look at page 1481 Where, using VARMA-GARCH and VARMA-AGARCH they have estimated parameters for Spillover effects α and β for Tourism Growth and Country risk return . From table 1 their dependent variable Country risk return of Cyprus θ1 and θ2 is constant and coefficient for conditional mean ,conditional variance own effects of CRR of Cyprus .where ω constant α is ARCH term and β Garch term .For Spillover effects α of TG of Cyprus and β of tourism growth of Cyprus and αTG_Malta β TG_Malta , α CRR_Malta ,β CRR_M .
How can I get this result using my four variable. ? is there and code I can use ?
Please help me
Kind Regards
rubel

Isn't that just GARCH with MV=CC,VARIANCES=VARMA or MV=CC,VARIANCES=VARMA,ASYMMETRIC applied to a model with each variable having a constant and one own lag?

Sir,
You are right ,i tried to do it in rats but could not figure out how to get this result .I did the followings:
1.Loaded the data on Rats.
2.Clicked on Time Series
3.Clicked ARCH/GARCH(Multivariate)
4.From the Multivariate Garch Wizard , I choose dependent variable as Country risk return of UK
Mean Model Variable just Constant and Variance Shift Varibles i choose Tourism growth rate of UK , Country risk return of France and Tourism Growth rate of France .
5. on Model Type i choose CC and Univariate Models I choose VARMA and Tick on Robust Standard errors ,Distribution i clicked on Normal , estimation method BHHH and tick on Asymetric effect for VARMA-AGARCH . and i got this result :

Usable Observations 216
Log Likelihood -462.0897

Variable Coeff Std Error T-Stat Signif
************************************************************************************
1. Mean(UKCRR) 68.61550030 0.12958032 529.52101 0.00000000
2. C 20.42043315 5.58768760 3.65454 0.00025764
3. A 0.33081877 0.20192489 1.63833 0.10135374
4. B 0.57188787 0.11628263 4.91808 0.00000087
5. UKTourism -0.00050681 0.00036100 -1.40391 0.16034666
6. FRNCRR -0.27122397 0.07878480 -3.44259 0.00057617
7. FRNTourism 0.00001351 0.00006065 0.22267 0.82379120

I need to find Theta 1 and 2 ,constant and coffeciant of AR(1) and Spillovers parameter Alpha and Beta like the McAleer(2007) paper , how can i get it ?

You have always shared the treasure of knowledge you have with me please share this time as well .I truly appreciate your kind help .

Kind regards

Rubel

You can't use the wizard to do that because the mean model is different from equation to equation. Just set up a model with separate AR equations for each of the variables as in the GARCHMV.RPF program:

*
* Univariate AR(1) mean models for each series, DCC model for the variance
*
equation(constant) jpneq xjpn 1
equation(constant) fraeq xfra 1
equation(constant) suieq xsui 1
group ar1 jpneq fraeq suieq
garch(p=1,q=1,model=ar1,mv=dcc,pmethod=simplex,piter=10)

You would just change this to fit your series and change to MV=CC and add VARIANCES=VARMA or VARIANCES=VARMA and ASYMMETRIC options.

BTW, it appears they suppressed the "other" asymmetry terms in the VARMA-AGARCH, so there will apparently be extra coefficients in your RATS output.

Sir,
I hope you are well and sound.I have implemented VARMA-GARCH in rats but not sure if i have correctly put the parameter
in right place .

i have also put the model(equation) i wanted RATS to do for me .

from rats output i have taken the value and put it on excel to see a clear picture .

Please take a look and please suggest if i am right or wrong .I have attached all the file .

Thank you for your continuous support and guidance.

Kind regards
Rubel

I have no idea what your four character codes mean. For both the A's and B's, i,j means the effect of j on i. A is for the "ARCH" (lagged squared residuals) and B is for the "GARCH" (lagged variance).

BTW, there's a reason the header is being SHOUTED. Your model isn't converged and probably isn't close to being converged given that it's stopping at iteration 37. You also really don't have much data for that complicated a model. And GARCH effects are often rather weak when you work with monthly data.


MV-CC GARCH with VARMA Variances - Estimation by BFGS
NO CONVERGENCE IN 37 ITERATIONS
LAST CRITERION WAS 0.0000000
SUBITERATIONS LIMIT EXCEEDED.
ESTIMATION POSSIBLY HAS STALLED OR MACHINE ROUNDOFF IS MAKING FURTHER PROGRESS DIFFICULT
TRY HIGHER SUBITERATIONS LIMIT, TIGHTER CVCRIT, DIFFERENT SETTING FOR EXACTLINE OR ALPHA ON NLPAR
RESTARTING ESTIMATION FROM LAST ESTIMATES OR DIFFERENT INITIAL GUESSES MIGHT ALSO WORK
Monthly Data From 1997:03 To 2007:12
Usable Observations 130
Log Likelihood -1300.0021

Sir,
Thank you so very much for your kind help.I have added 7 more years monthly data but there is still problem with convergence is it close to converged ? Is there any solution ?

MV-CC GARCH with VARMA Variances - Estimation by BFGS
NO CONVERGENCE IN 200 ITERATIONS
LAST CRITERION WAS 0.0000739
Monthly Data From 1997:03 To 2014:12
Usable Observations 214
Log Likelihood -2555.1009

Then i tried to increase iterations=450 and i got this :

MV-CC GARCH with VARMA Variances - Estimation by BFGS
NO CONVERGENCE IN 450 ITERATIONS
LAST CRITERION WAS 0.0000358
Monthly Data From 1997:03 To 2014:12
Usable Observations 214
Log Likelihood -2554.7362

Thank you very much for your kind help.

Kind regards
rubel

It's stopped because it's hit the iterations limit, so keep increasing it. (I would probably just put in 2000 now). The log likelihood isn't changing that much and the criterion is close to zero, so it's likely rather close to convergence.