how can I estimate a dynamic model?

[samuelillo89]

hello, i am quite new in using Eviews, and I have been asked to do a project with economic variable, for which, within other staff, i have to estimate a dynamic model regarding one variable( in my case the exchangerate of a country) i am stuck on that step and i was wondering if someone could help me over here..

thanks a lot!

[EViews Gareth]

Could you define what you mean by "dynamic model"?

[samuelillo89]

i am asked in the previous sections to choose the best univariate model for each variable(exchange rate) i did that using quick-estimate equation- and trying different ARMA models then i had to evaluate the best one using BIC criterion, then (where i am stuck) i am required to "estimate a dynamic model for each variable(the other is gdp)" and then calculate the impact multipliers, median lag and test for existence of granger casualty between the two variables. later on, they ask me if the dynamic model predicted is better than the univariate model i found before..

That is all I can tell you, i hope there is enough information, anyway, i don't know how many types of dynamic models exist (as i said, this is my first time with eviews) but it must be something afordable to beginners .

thank you

[EViews Gareth]

I'm afraid that term is simply too general to be of any use. The ARMA models you have already tried could be said to be a dynamic model.

[samuelillo89]

If I estimate an arma(4,8) for my variable using all the observations except the last 20, and then forecast using dynamic forecast with all the observations, do you think that would be ok?

[samuelillo89]

I'm sorry, forget about the 20 observations, just estimate a dynamic forecast from my arma model

[samuelillo89]

sorry to cother again, but i have another question, how can i decide which model is better based on the prediction(for the same variable)? which indicator do i have to focus in?