Help: Test of weak instrument after Heckman along with IV estimation

[Jean-Louis BAGO]

Hi Dear Statalist,

I'm using a control function with Heckman to correct for endogeneity and sample selection in my model:
Y is the dependent variable, X is the endogenous count data variable, Z the instrument, S the selection variable and R a vector of variable.

Here is the stata code:

glm X Z $R, family(poisson) link(log)
predict res, l
heckman Y X $R res , select(S= X $R res ) vce(r)

Now my question is how to test for weak identification in this estimation?

Any help will be greatly appreciate.

Best regards,

Jean-Louis

[Phil Bromiley]

You'll increase your chances of a useful answer by following the FAQ on asking questions - provide Stata code in code delimiters, readable Stata output, and sample data using dataex.

I don't know how to handle the weak identification problem in a control function model, and I'm not sure that you have to use a control function approach - heckman may be consistent even with an endogenous count variable. Also, the eregress set of commands are made for this kind of problem.