PDL Lag Selection

[weiwensg]

Hi everyone,

I am estimating a panel data fixed-effects model in Eviews 7.

Y is the dependent variable. My independent variables are X and a PDL (polynomial distributed lag) of Z.

Z has intermittent missing data.

I would like to choose the PDL of Z which provides the best fit. I have designed an automated program to generate many different models using different PDL parameters. However, the problem is that the goodness of fit measures are not comparable. The models which use more lags have fewer observations than those with less lags.

Is there a way to use the same observations for each regression?

(I understand that there is a "laglen" command when estimating VAR models; I am looking for an analogous command.)

[weiwensg]

Also, if there are any model selection commands for PDLs, do let me know.