Hi,
I am dealing with the well-known problem of health costs as my DV which leads to zero-inflation and heteroscedasticity. Consequently, I am estimating a two-part model with my first part being a random effects probit model (Pr(Y >0| x)) and my second part is a random effects OLS model with E (ln(y)| y >0, x).
dum_Y = 1 if Y > 0 otherwise it is 0.
xtprobit dum_Y $X
xtreg ln(Y) $X if Y > 0
Now, I want to estimate the marginal effects based on both parts. Unfortunately, I can not apply the twopm command by Belotti et al. (2015) since I want estimate random effects probit and OLS models in both parts.
Can someone help me how to estimate those marginal effects manually in stata.
Thank you.
