Lagging variables in instrumental variables regression ivreg2

[Veeresh Thummadi]

Hi I would like to know how I can lag variables in the instrumental variable regression to estimate the effect of education on wages. Below is the example from widely used example of the effect of parents income on wages. As I have already lagged the education should I lag the fathers income(fincome) by 2 lag periods or 1 to overcome simultaneity bias, Please suggest if I am doing this correctly. Below is my code.

xtset year id

ivreg2 lnwages L.union (L.educ= L.fincome i.year i.state),  sfirst savefirst robust cluster(id)

Thanks

[Phil Bromiley]

The issue is whether the iv's are exogenous. Sometimes, we get that by lags, but sometimes by theory. Whether a specific variable should be lagged more depends on your theory and understanding of the specific context.