Dear all
I analyses the child’s probability to work and go to school.
One important explanatory variable is income, which is endogenous. To correct for this I use an instrument. My baseline model therefore look like this:
y*1i = β1x1i + β2x2i + β3y3i +ε1i y1i=1 if y*1i > 0, otherwise 0
y*2i = β3x1i + β4x2i + β3y3i + ε2i y2i=1 if y*2i > 0, otherwise 0
y3i = β3x1i + β4x2i + β5x5i +ε3i
I estimate my model using the CMP program (Roodman 2011)::
cmp (exp = x1 x2 x3 inst) (work = exp x1 x2 x3) (school = exp x1 x2 x3) if age>5 & age<15, cluster(district) ind($cmp_cont $cmp_probit $cmp_probit)
After this I estimate the marginal effects, by using :
margins, predict(pr eq(#2)) dydx(*) force
However, now I want to include random effects at household and district level, changing my model into:
cmp (exp = x1 x2 x3 inst ||district: || hhold
(work = exp x1 x2 x3 ||district: || hhold (school = exp x1 x2 x3 ||district: || hhold if age>5 & age<15, cluster(district) ind($cmp_cont $cmp_probit $cmp_probit)
My question is: How do I estimate the marginal effects after this model?
Thankfull for any help!
//Elin Vimefall
Ref
Roodman D. (2011). Fitting fully observed recursive mixed-process models with cmp. The Stata Journal 2011, 11(2), pp 159-206.
I analyses the child’s probability to work and go to school.
One important explanatory variable is income, which is endogenous. To correct for this I use an instrument. My baseline model therefore look like this:
y*1i = β1x1i + β2x2i + β3y3i +ε1i y1i=1 if y*1i > 0, otherwise 0
y*2i = β3x1i + β4x2i + β3y3i + ε2i y2i=1 if y*2i > 0, otherwise 0
y3i = β3x1i + β4x2i + β5x5i +ε3i
I estimate my model using the CMP program (Roodman 2011)::
cmp (exp = x1 x2 x3 inst) (work = exp x1 x2 x3) (school = exp x1 x2 x3) if age>5 & age<15, cluster(district) ind($cmp_cont $cmp_probit $cmp_probit)
After this I estimate the marginal effects, by using :
margins, predict(pr eq(#2)) dydx(*) force
However, now I want to include random effects at household and district level, changing my model into:
cmp (exp = x1 x2 x3 inst ||district: || hhold
(work = exp x1 x2 x3 ||district: || hhold (school = exp x1 x2 x3 ||district: || hhold if age>5 & age<15, cluster(district) ind($cmp_cont $cmp_probit $cmp_probit)My question is: How do I estimate the marginal effects after this model?
Thankfull for any help!
//Elin Vimefall
Ref
Roodman D. (2011). Fitting fully observed recursive mixed-process models with cmp. The Stata Journal 2011, 11(2), pp 159-206.
Comment