gmm xtabond2

[Giorgio Carlini]

I'm estimating a dynamic panel model (n=20 and T=15) with GMM system. The dependent varaible is the rate of growth and the main independent variable is migration flow.
In this type of model to control for the source of endogeneity of migration, this is generally intrumented with a shift-share instrument.
I was wondering if I can use such instrument in a dynamic GMM estimation given that this procedure uses internal lagged instruments.
thanks

[George Ford]

xtabond2 is designed for large N, small T. You don't have large N (I think rule of thumb is about 30 min, but keep instruments to a min). I think you can use xtabond2 with just external instruments (setting the shift share as external IV).

[Giorgio Carlini]

thanks a lot George