Endogenous variable

#1
Hi,

I am implementing a two-stage OLS regression to account for the endogeneity of a variable. My instruments are time invariant. However, the second-stage regression requires yearly dummy variables. Would it be possible to include the year dummies in the second stage model but exclude them from the first stage model?

It doesn't seem like there is a lot of literature on this...

Thanks very much
 
#2
I am not sure what you mean by "the second-stage regression requires yearly dummy variables." Do you mean that there are unobserved year effects that are not controlled for by your other variables? Would this not be true in the second stage? I think it all depends on what your first stage ind. var. is and what you are modeling.

If your ind. var. is time variant, another option is to difference the equation. But again, it all depends on what you are trying to model.