Riccardo Jack Lucchetti r.lucchetti at univpm.it
Fri Feb 17 13:14:23 EST 2006

On Fri, February 17, 2006 19:02, ohinata manabu wrote:
> I am currently using Gretl to run GARCH model for volatility of daily
> return of IBM stock estimation.
> However, I can't match the result from gretl and the GARCH formula.
>
> What I know is: ht=$B&X!\&A!J#r!]#m!K(B^2$B!\&B&R(B^2
>
> But, what I got is:
>
> Variable     Coefficient
> const        0.000553614
> alpha(0)     9.90527e-06
> alpha(1)     0.108505
> beta(1)      0.879647
> Mean of dependent variable = 9.79994e-005
> Standard deviation of dep. var. = 0.0244575
> Unconditional error variance = 0.000836019
> Log-likelihood = 2387.036
> AIC = -4764.073
> BIC = -4739.534
>
> I think $B&A(B(0) is $B&X(B.
> But I don't know which is ht and what does const mean.
>
> Thank you for your help.
> Have a good weekend.

By default gretl estimates a model where the conditional mean is constant, so
"const" is the estimate of m in the following pair of equations:

y_t = m + e_t
V(e_t) = h_t = alpha_0 + alpha_1 * e_{t-1}^2 + \beta_1 h_{t-1}

note that V(e_t) should be read as the *conditional* variance of e_t.

In your example, the estimate for m is 0.000553614 and so on.

--
Riccardo "Jack" Lucchetti
Dipartimento di Economia
Facoltà di Economia "G. Fuà"
Ancona