Dear all,

I estimate 3 bivariate probit models (for three countries), using biprobit command with robust st. errors and survey weights. I estimate marginal probabilities, p11, p10, p01, p00. Is it possible to test if there are differences in marginal effects between countries for every explanatory variable? I can use command test for testing equality of coefficients, but not equality of marginal effects.

This is the command for testing equality of coeff. between equations.
biprobit (y1$xlist1) (y2=$xlist1) ///
[iw=weight] if country==1, nonrtol // I run this three times
estimate store x
suest x1 x2 x3
test [x1_y2]age = [x2_y2]age = [x3_y2]age, mtest(b)

I use this command to save marginal effects, but I don't know how to test equality of marginal effects betwen three countries.

forvalues i=0(1)1{
forvalues j=0(1)1{
biprobit (y1=$xlist1) (y2=$xlist1) ///
[w=weight] if country==x, vce(r) nonrtol
margins, dydx(*) predict(p`i'`j') post
estimate store x`i'`j'
}
}

Thank you in advance.

Best,
Aleksandra