When I use cmp_probit estimation model, the tool variable coefficient exists, but the tool variable is omitted when marginal effect is used. Is this because of severe multicollinearity? Is there a way to test multicollinearity for cmp commands?
the code:
cmp (y=x1 x2 x3 x4 x5) (x1=x2 x3 x4 x5 z) , ind($cmp_probit $cmp_probit) tech(dfp) nolr quietly robust
margins, predict (pr eq(#1)) dydx (*) force