Multinomial probit instrumented with cmp

[Jim Smith]

Hello,

I would like some information about the cmp command.

My issue is the following.

I want to instrument a multinomial probit model.

My model looks like this: cmp (Y = X1 X2 ... ENDOGENEOUS, iia), nolr ind ($ cmp_mprobit) vce (robust)

My endogenous variable is a score from 0 to 6. My instrumental variable is dichotomous. Here is the code I made to instrumentalize my model:

cmp (Y = X1 X2 ... ENDOGENEOUS, iia) (ENDOGENEOUS = X1 X2 ... INSTRUMENT, iia), nolr ind ($ cmp_mprobit $ cmp_cont) vce (robust)


Nevertheless, my ENDOGENEOUS variable is a score variable, so it makes more sense to write $ cmp_oprobit instead of $ cmp_cont.

Is this formula right to instrument my model ? :

cmp (Y = X1 X2 ... ENDOGENEOUS, iia) (ENDOGENEOUS = X1 X2 ... INSTRUMENT), nolr ind ($ cmp_mprobit $ cmp_oprobit) vce (robust)

Indeed, I want to be sure that the integration of this ordered variable instrumentalizes my model.

Finally, I try to test the validity of my instrument with the command "overid", however stata returns me an error message. Do you have a solution for a multinomial probit model?

Thank you.

Jim

[Phil Bromiley]

You didn't get a quick answer. 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. It is important that we be able to run your model and that we know exactly what you ran and what Stata gave you. You can simplify your model to the minimum necessary to generate the problem.

cmp is user written. You may need to contact the author about your questions. You can probably do this with the eregress routines in Stata 15 or GSEM.

[Jim Smith]

Hello,

Thank you very much for your advice.