Huge (dummy) question with MA processes

[CatarinaSilva]

Hi1

I'm revising this model:

Y(t) = c + b1*AR(1) + b2*MA(2) + b3*MA(3) + b4*MA(4) + b5*MA(5)

My question is: can a model have more than one MA process at the same time? I realized that the r-squared is higher if we account for more than one MA, but does it have a statistical meaning? It works but, is it right or it is just a spurious model? I have the results in attachment.
Thank you so much.

[startz]

This isn't more than 1 MA process, it's a single MA(5) process.

[CatarinaSilva]

Ok, so the right specification would be Y(t) = c + b1*AR(1) + b2*MA(5)

But, as you can see in the image, if I estimate this equation, the MA(5) is non significant.

[startz]

Sorry, EViews notation for the order of an MA process is different from the usual statistical notation. What's normally called an MA(5) process needs to be written in EViews as

Code: Select all

ls y c ma(1 to 5)

[CatarinaSilva]

I did'nt know that. Thank you very much.