Dear all,

I'm applying the control function (CF) approach in a nonlinear model, the fractional response model, to control for two potentially endogenous variables. I know that when using 2SLS to address endogeneity in linear models, the weak instrument issue is critical. However, in materials for the control function approach, I hardly found any discussions regarding the weak instrument issue. Does it not matter for CF approach?

I tried to take advantage of the -ivregress 2sls- command and looked at the first stage F statistics. It shows that for one endogenous variable the F-statistic is below 10, while for the other one it is greater than 10. Would it have any consequence to my estimation, as it could have in 2SLS estimations?

Any help would be greatly appreciated!

Best,
Ziwei