Hi,

I am doing a fixed effect panel data regression (T=40 N=1087).

Theoretically my independent variable must be lagged at least one year (4 quarters), possibly more.

I get statistical significant variables when I lag 4, 8 12 and 16 quarters. How do I know which lag to present/use?
Is it best to use the highest or the lowest lag? Can I conduct some kind of tests?

The only thing that changes a little is the control variables and the R squared which reduce a little for each higher lag.

I would like to decide on a specific lag, before doing several small alternations to the model.

Thank you in advance.