Interpret ATE from logs linear model

[Nikolaos Kanellopoulos]

Hi everyone,

I am using eregress to estimate a linear regression with endogenous treatment where the dependent variable is in logs, mostly to satisfy the normality assumption [see code below]. I have a couple of questions regarding the results:

1. What is the appropriate interpretation of the ATE?
2. Can the ATE be transformed into levels?

Thank you in advance.

use http://www.stata-press.com/data/r15/wageed
gen lnwage = ln(wage)
qui eregress wage c.age##c.age tenure, entreat(college = i.peduc) vce(robust)
estat teffects, ate
qui eregress lnwage c.age##c.age tenure, entreat(college = i.peduc) vce(robust)
estat teffects, ate