Calculating inverse Mill's ratio
Posted: Wed Mar 03, 2010 1:07 am
Hi guys,
I'll greatly appreciate if someone could help me with the following. I've used Eviews mostly for OLS/GMM estimations and never used it for probit/logit analysis. Now I came to the point where my research requires estimation using the so-called "treatment effect" methodology. That is, I first need to estimate the "decision" equation where the dependent variable is a binary one, using probit. Then I have to take the inverse Mill's ratio from the estimation of the first probit equation and put it in the second (OLS) regression as an additional explanatory variable. I don't know how to calculate the Inverse Mills ratio. However, I found the article where the auhtors say that "after estimating a probit model, Eviews stores the inverse Mills' ratio for each observation under the series named RESID". I wonder, is it that simple? Does Eveiws really calculate the ratio automatically and stores it for each observation as a Resid series? Thanks a lot!!!
I'll greatly appreciate if someone could help me with the following. I've used Eviews mostly for OLS/GMM estimations and never used it for probit/logit analysis. Now I came to the point where my research requires estimation using the so-called "treatment effect" methodology. That is, I first need to estimate the "decision" equation where the dependent variable is a binary one, using probit. Then I have to take the inverse Mill's ratio from the estimation of the first probit equation and put it in the second (OLS) regression as an additional explanatory variable. I don't know how to calculate the Inverse Mills ratio. However, I found the article where the auhtors say that "after estimating a probit model, Eviews stores the inverse Mills' ratio for each observation under the series named RESID". I wonder, is it that simple? Does Eveiws really calculate the ratio automatically and stores it for each observation as a Resid series? Thanks a lot!!!