Panel data mixed effects with IV

[Ben Lambert]

I would like to estimate a model that allows for mixed effects in a slope coefficient on an endogenous variable. I know I can do,

xtmixed y x || id: x

to estimate a model which allows varying intercepts on the x variable. I do not know, however, how to have a mixed model when x is endogenous. I have an instrumental variable z, and so can get consistent estimates of the fixed effect slope parameter, but do not know how to estimate a models where I allow the parameter to vary.

Does anyone know how to do this?

An added complexity, which I will probably need to tackle later is that the model contains a lagged dependent variable. And so, really, I would be wanting to extend Arellano-Bond/Bover to include a mixed effects coefficient. Again, any ideas on how to do this are most welcome!

[Giovanni Colitti]

Hi Ben,

I have the exact same question, including the added complexity of a lagged dependent variable--which I am also content with tackling later. Did you find a solution to your question? Does anyone know how to account for endogeneity in one of the independent variables in a mixed effects model framework?

[Oliva Safari]

Hello, did you find solution?