IV instrument with contemporaneous and lagged values

[Giorgio Carlini]

Hello,
I'm estimating a panel fe model using instrumental variable
I've instrumented my endogenous variable X(t-1) with Z(t-1)
However, I've also used Z(t-1) togheter with Z(t-2), and estimation results are much better.
I'm not sure that this is fine. Are there any problems of including Z(t-2) too?
Thanks

[Joro Kolev]

The validity of instruments is determined from outside of the model, from prior reasoning.

Hence this is not a question that is possible to answer just on the basis of the precedence information you provide.

[Giorgio Carlini]

Hello Joro,
sorry I was not clear. The economic justification from using Z as instrument is reasonable. Also the statsistical tests show that the instruments Z(t-1) and Z(t-2) are valid when used togherer.
And, as I said, when using both estimation results are much better.
So my only worried is if they can be used simultaneously
Thanks

[Joro Kolev]

Yes, of course they can be used simultaneously, if you think that both of them taken separately are valid instruments.

[Giorgio Carlini]

Thanks a lot Joro,
really appreciated your help