STATA's truncreg command comes with a marginal option that prints out the marginal effect of the independent variable(s) on the dependent variable. This option assumes that the independent variable is continuous. When the independent variable is discrete, the STATA manual suggests using the following postestimation command:

mfx compute, predict(e(

*a*,

*b*))

where

*a*and

*b*are the lower and the upper limits on the dependent variable.

When I use this postestimation command following the truncated regression of a continuous variable (say, y) on a discrete (binary, or dummy) variable (say, x), the marginal effect that STATA prints out is identical to the OLS coefficient I obtain from "regress y x." I tried this for different continuous dependent variables, and observed it happening on a regular basis. So it doesn't look like it is a coincidence.

My question is: what is the explanation?

Any and all help will be appreciated.