biprobit and instuments

[Elin Vimefall]

Hi
I'm analyzing the child’s probability to be working together with the decision to be in school using a bivariate probit model:

biprobit work school x1 x2 x3

However one of my explanatory variables turns is endogenous for the work decision. To correct for this I want to use an instrument. Is there any way to do this while still using the biprobit?

Best regards /Elin Vimefall

[Stephen Jenkins]

Suppose it's "x1" that's endogenous. You could add an equation for x1 to your system and allow the errors in your 3-equation system to be correlated. (You also need instruments or other conditions satisfied for Identification, of course.) Have a look at cmp on SSC.

[Elin Vimefall]

Thank you Stephen! Does it matter that x1 only is endogenous in one of the equations?

//Elin

[Stephen Jenkins]

No (but I'm guessing -- I am not an expert on this).