Margins after Recursive Bivariate Probit Model

[naveed ahmed]

Dear Statalisters,

I am trying to run the following model in stata 15

y=binary variable
b=binary variable
a=continuous variable
c=binary variable

biprobit (eq1:Y= a b c) (eq2:b=a c)

After running this command I tried obtaining marginal effects using the following command:

margins c, predict(equation(eq1))

Stata reports the following: prediction is a function of possibly stochastic quantities other than e(b),r(498)

Any advice on how to estimate margins after running a recursive bivariate probit will be highly appreciated.


Regards,
Naveed