ARMA model and Excel

[Kvitko]

Hi,

I am trying to manually count an ARMA model forecast in Excel, but I am experiencing few issues. The model equations looks like this:

OLS: Y = c + X + AR term + MA term

I have actual forecast from EVIEWS, but I am not able to get same results in Excel. So I am kindly asking for any help/advice/guide, how to properly do it. In Excel I just follow the equation and use estimated coefficients.

Thanks in advance!

[KrilleJ]

Hard to say with limited information.

Have you imported estiamted disturbances into Excel?

Have you specified AR-terms in EViews as a lagged dependent variable or as an AR-form for the error term?

/K

[Kvitko]

I have AR/MA-form for the error term.

Estimated coefficients looks like this. All are significant, used HAC estimate because of heteroscedasticity issue and serial correlation.

d(kurz_sa) = @coef(1) + @coef(2) * d(bilance1_sa) + [ar(7) = @coef(3) , ma(1) = @coef(4) , backcast = 2005m01 , estsmpl = "2005M01 2013M10"]

@coef(1) = -0.0437984
@coef(2) = 0.6217076
@coef(3) = -0.3362004
@coef(4) = 0.2313841

Estimated S.E. = 0.3441704

[EViews Gareth]

Obvious question, but have you looked at the example?
http://forums.eviews.com/viewtopic.php?f=7&t=465

[Kvitko]

Thanks Gareth! Went through it and was able to replicate static forecast. No I will try dynamic. Helped me a lot.