Unbalanced Panel fixed Effect

[sweivE]

Hi,

I am using eviews 7 and I am trying to estimate an unbalnced panel with period and cross-section fixed effects. my specification is nonlinear as follows:

dlog(y)=c(2)*(log(y(-1))+(c(3)/c(2))*log(xk)+(c(4)/c(2))*log(z)+c(1))+c(5)*dlog(x)+c(6)*dlog(z)

I get the following error message:

"Explicit two-way effects specifications not allowed with unbalanced date"

Is there a way to bypass this problem?

Thanks!

[EViews Glenn]

You can do a one-way fixed effects (using your largest dimension), and then add dummy variables for the other dimension.

[sweivE]

Thanks Glenn!

Here's another one for you:

If instead of dynamic fixed effect I would like to use a 'pooled mean estimator'? Is that option available in eviews? e.g. something similar to 'xtpmg' command that can be used in STATA?

[EViews Glenn]

Unfortunately, not as a built-in feature. It's on our list of things to look at...