Instrumental variable analysis

Dear people,

In my simple regression analysis are two variables endogenous.
GPAi = α + β•Wi +β•Wi^2 + θ•Si + θ•Si^2 + Qi•φ + εi
Where Wi and Si are correlated with the error term (and therefore endogenous) in this simple OLS regression

I want to do Two Stage Least Squares. However, I have a few questions:

- Do I need to check again whether the quadratic terms of Si (studyhours2) and Wi (workhours2) are necessary? (after 2SLS by the command ovtest maybe?)

- Do I also need to instrument Wi^2 and Si^2 ?? Or is (at least) two instruments for Wi and Si enough?

- If I want to check whether my instrument has enough correlation with the endogenous regressor by looking at the F-statistic (>10). How should I do this? estat firststage does not work (stata 10 gives : invalid subcommand firststage) Is there a command that works?
(reg workhours nodisability unearninc grade workhours2 studyhours2 age lifeaway univers prevgrade motivation if stuyear==1, for example ?)