## ARMA(2,2) - GARCH-M estimation

For technical questions regarding estimation of single equations, systems, VARs, Factor analysis and State Space Models in EViews. General econometric questions and advice should go in the Econometric Discussions forum.

Moderators: EViews Gareth, EViews Moderator

ecofine
Posts: 28
Joined: Fri Sep 26, 2014 2:24 am

### ARMA(2,2) - GARCH-M estimation

Dear
I am using version 10 of Eviews.
I used GARCH-M estiamtion method and tried to estimate the volatilty model of stock returns.
But I got this kind of error messages.
Could you kindly le tme know what happens in the covariance?
The main equation for stock returns is ARMA (2,2) with constant.
And run GARCH-M (1,1). then I got this errors.
-----------------------------------------------------------------------------------------
Failure to improve likelihood (non-zero gradients) after 1 iteration
Coefficient covariance computed using outer product of gradients
WARNING: Singular covariance - coefficients are not unique
-----------------------------------------------------------------------------------------
So, all the resutls are nonavailable.
Presample variance: backcast (parameter = 0.7)

GARCH = C(7) + C(8)*RESID(-1)^2 + C(9)*GARCH(-1)

Variable Coefficient Std. Error z-Statistic Prob.

@SQRT(GARCH) 3.900948 NA NA NA
C -0.092105 NA NA NA
AR(1) 0.005000 NA NA NA
AR(2) 0.005000 NA NA NA
MA(1) 0.005000 NA NA NA
MA(2) 0.005000 NA NA NA
Variance Equation

C 0.000365 NA NA NA
RESID(-1)^2 0.150000 NA NA NA
GARCH(-1) 0.600000 NA NA NA

Mean dependent var 0.000314 S.D. dependent var 0.023701

Inverted AR Roots .07 -.07
Inverted MA Roots -.00-.07i -.00+.07i

-----------------------

May i get any commenst on this results?

Thank you very much for your kind and heloful information in advance.

HIS

startz
Non-normality and collinearity are NOT problems!
Posts: 3337
Joined: Wed Sep 17, 2008 2:25 pm

### Re: ARMA(2,2) - GARCH-M estimation

Try different starting values in the C vector.

Also, why would you think that ARMA(2,2) is a sensible model for stock returns? (I'm assuming that the data is from a more or less efficient market.)

ecofine
Posts: 28
Joined: Fri Sep 26, 2014 2:24 am

### Re: ARMA(2,2) - GARCH-M estimation

Thank you, Starz,

I found that a stock returns could be fitted by ARMA model and determined.
By the way, what is C vector, do you mean covariance vector?
Could you kindly introduce me how to change C-vector?

Thank you very much!

HIS

startz
Non-normality and collinearity are NOT problems!
Posts: 3337
Joined: Wed Sep 17, 2008 2:25 pm

### Re: ARMA(2,2) - GARCH-M estimation

The C vector is the coefficient vector. It appears in the workfile and you can open and edit it like any other object.

I am curious why you think an ARMA(2,2) model is appropriate, and what stock market returns you are using.