Dear,
I am trying to apply marginal effects after Tobit II (Heckman two-step procedure for correcting the selection bias).
The get marginal effects from fist, selection equation I can use these two commands: margins, dydx(*) predict(psel)
mfx compute, predict (psel)
To get marginal effect of the second, outcome equation, these two commands are available:
mfx, predict(yexpected)
margins, dydx(*) predict(yexpected)
The issue is that after the second sets of commands (marginal effects of outcome equations), variables included in the 1st stage (selection equation) appear in the 2nd stage (outcome equation) even though they were not included when Heckman model is applied per se.
Does anyone know how I can have marginal effects overlap with the variable actually included in the 2nd stage? In other words, how can I prevent variables from the 1st stage appearing in the second stage when marginal effects are applied?
Thank you.
I am trying to apply marginal effects after Tobit II (Heckman two-step procedure for correcting the selection bias).
The get marginal effects from fist, selection equation I can use these two commands: margins, dydx(*) predict(psel)
mfx compute, predict (psel)
To get marginal effect of the second, outcome equation, these two commands are available:
mfx, predict(yexpected)
margins, dydx(*) predict(yexpected)
The issue is that after the second sets of commands (marginal effects of outcome equations), variables included in the 1st stage (selection equation) appear in the 2nd stage (outcome equation) even though they were not included when Heckman model is applied per se.
Does anyone know how I can have marginal effects overlap with the variable actually included in the 2nd stage? In other words, how can I prevent variables from the 1st stage appearing in the second stage when marginal effects are applied?
Thank you.
Comment